Structural Hierarchy of Regulatory Elements in the Folding and Transport of an Intestinal Multidomain Protein*

Human intestinal lactase-phlorizin hydrolase, LPH, encompasses four homologous domains, which presumably have evolved from two subsequent duplications of one ancestral gene. The profragment, LPHα, comprises homologous domains I and II and functions as an intramolecular chaperone in the context of the brush-border LPHβ region of LPH. Here, we analyze the inter-relationship between homologous domains III and IV of LPHβ and their implication in the overall structure, function, and trafficking of LPH. In silico analyses revealed potential domain boundaries for these domains as a basis for loop-out mutagenesis and construction of deletion or individual domain forms of LPH. Removal of domain IV, which contains lactase, results in a diminished phlorizin hydrolase activity, lack of dimerization in the endoplasmic reticulum (ER), but accelerated transport kinetics from the ER to the Golgi apparatus. By contrast, deletion of domain III, which harbors phlorizin hydrolase, generates a malfolded protein that is blocked in the ER. Interestingly, homologous domain III is transport-competent per se and sorted to the apical membrane in polarized Madin-Darby canine kidney cells. Nevertheless, it neither dimerizes nor acquires complete phlorizin hydrolase activity. Our data present a hierarchical model of LPH in which the homologous domain III constitutes (i) a fully autonomous core domain within LPH and (ii) another intramolecular chaperone besides the profragment LPHα. Nevertheless, the regulation of the trafficking kinetics and activity of domain III and entire LPH including elevation of the enzymatic activities require the correct dimerization of LPH in the ER, an event that is accomplished by the non-autonomous domain IV.

Construction of cDNA Clones-Deletion mutants of intestinal LPH were generated by loop-out mutagenesis PCR with the plasmids pcDNA3-LPH, pLPH-GFP (24), and pLPH-YFP as templates. LPH domains III and IV share a comparable, quite high degree of sequence identity (around 40%) with highly conserved family 1 glycoside hydrolases of known three-dimensional structure from humans (25), plants (26), fungi (27), and bacteria (28,29). Structure-based sequence alignments with the aforesaid homologous proteins generated by GenTHREADER (30) were used to dissect the LPH domains. Structural alignments were also used to build models of domains III and IV using SWISS-MODEL (31). pcDNA3-LPH was generated by cloning the wild type LPH cDNA in the vector pcDNA3 (Invitrogen) with the EcoRI sites of pLPH (32). LPH was fused to YFP by subcloning the EcoRI/ScaI fragment from pcDNA3-LPH containing full-length LPH in-frame into the EcoRI/SmaI-digested pEYFP-N1 vector (Clontech-Takara, Saint-Germain-en-Laye, France) to create pLPH-YFP. A construct comprising the signal sequence and homologous domain III (D3) of LPH was made by inserting a stop codon into the plasmid pJB20-LPH␤final with the same system generating plasmid pD3. For the generation of a GFP fusion protein, the cDNA encoding the signal sequence of LPH and homologous domain III was amplified by PCR with pD3 as template and cloned into the EcoRI/AgeI-digested pEGFP-N1 vector (Clontech-Takara). The applied oligonucleotides were purchased from Sigma and are listed in Table 1.
Transient Transfection of COS-1 Cells, Metabolic Labeling of Cells, Immunoprecipitation of Cell Extracts, and SDS-PAGE-COS-1 cells were transiently transfected with DNA using DEAE-dextran essentially as described previously (32). The cells were biosynthetically labeled with 80 Ci of [ 35 S]methionine in methionine-free minimum essential medium. Labeling was performed either continuously or following a pulsechase protocol where the labeled cells were chased with nonradioactive methionine for different periods of time. In some experiments, the biosynthetically labeled cells were washed with Dulbecco's modified Eagle's medium and treated with 50 l/ml trypsin for 30 min at 37°C as described previously by Naim et al. (32). This treatment is intended to probe for cell surface expression of wild type LPH and the deletion mutants. Immunoprecipitation of LPH or the deletion mutants from detergent extracts of the labeled cells was performed according to Naim et al. (12) using a mixture of mAb anti-LPH (HBB 1/909 and MLac1, MLac2, MLac4, MLac6, and MLac10), and V496. This mixture recognizes different conformations of LPH. Immunoprecipitates were treated with endo H or endo F (both from Roche Diagnostics, Mannheim, Germany) where indicated according to Naim et al. (12) followed by analysis using SDS-PAGE. The radioactively labeled protein bands were visualized by phosphorimaging and quantified with Quantity One software (Bio-Rad, Munich, Germany).
Transfection of MDCK Cells, Generation of a Stable MDCK-D3 Cell Line, and Analysis of Sorting-To generate stable MDCK cell lines expressing domain III (MDCK-D3), the cells were transfected using Metafectene Pro obtained by Biontex (Martinsried, Germany) following the manufacturer's instructions. Selection of stable cells was performed in the pres- ence of 0.3 mg/ml active G418 (Carl Roth GmbH), and after 14 -21 days, colonies were isolated and subcultured, and stable transformants were screened by immunoprecipitation. For the analysis of sorting behavior, MDCK-D3 cells were cultured on Transwell filters (obtained from Greiner) and biosynthetically labeled with [ 35 S]methionine. Proteins that have been secreted into the apical and the basolateral medium, respectively, were isolated by adding anti-LPH mAb and protein A-Sepharose to the removed media. Intracellular proteins were isolated by immunoprecipitation after cell lysis. The immunoprecipitates were analyzed by SDS-PAGE and phosphorimaging.
Trypsin Treatment of Immunoprecipitates-To assess the sensitivity of the mutants to trypsin, immunoprecipitated proteins were washed for an additional two times with phosphatebuffered saline containing 0.2% Triton X-100 and were supplemented with 10 g of bovine serum albumin as a carrier and incubated with 0.33 mg/ml trypsin for the indicated times at 37°C. The reaction was stopped by boiling for 5 min in SDS-PAGE sample buffer prior to gel electrophoresis.
Confocal Fluorescence Microscopy-COS-1 cells grown on coverslips were transfected, and confocal images of living cells were acquired 2 days after transfection on a Leica TCS SP2 microscope with a ϫ63 water Plan-Apochromat lens (Leica Microsystems, Bensheim, Germany) (34) and processed with the public domain ImageJ software package (available through the National Institutes of Health). For colocalization studies, the cells were co-transfected with GFP-or YFP-tagged LPH cDNA or the mutant LPH cDNA and the protein marker for the ER ER-DsRed and for the Golgi apparatus Golgi-CFP (Clontech-Takara).
Enzymatic Activity Assay-Lactase and phlorizin hydrolase activities were assessed as follows. The 35 S-labeled immunoprecipitates were washed with phosphate-buffered saline containing 0.2% Triton X-100 and incubated with 100 l of this buffer containing lactose or phlorizin at 28 mM final concentrations. The samples were incubated at 37°C for 2 h, and the amount of released glucose was assessed by high-performance liquid chromatography (HPLC). The quantification of the specific activity was related to the radioactive protein band detected by phosphorimaging and the number of methionines for each construct.

RESULTS
LPH comprises four extracellular regions that contain 38 -55% identical residues (1). An interval of about 100 amino acids within each domain is even more homologous, and this internal homology can also be found by comparison of LPH primary sequences of different species. Although the function of domains I and II that constitute the profragment or proregion of LPH, LPH␣, has been assessed before and shown to act as an intramolecular chaperone, the individual roles of the two other domains III and IV are poorly understood. The impact of these homologous domains on the generation of a transportcompetent configuration of LPH was addressed in conjunction with the question of whether either domain can fold independently. For this, we first generated several cDNA constructs, each lacking the coding region of one homologous domain (Fig.  1B). In silico analysis provided us with potential domain boundaries as a basis for site-directed loop-out mutagenesis. Fig. 1A shows main structural and functional features of human intestinal LPH. Fig. 1C presents homology-based models of domains III and IV showing typical TIM barrel structure of family 1 glycoside hydrolases.

Expression of Wild Type LPH and Domain Deletion Mutants in COS-1 Cells
To examine the contribution of each of the two homologous domains to the structural, functional, and trafficking features of LPH, the LPH deletion mutants were expressed in COS-1 cells, and their characteristics were compared with those of wild type LPH (Fig. 2). Cell lysates were immunoprecipitated, and the precipitated proteins were treated with endo H to determine their glycosylated state as a measure of trafficking capacity (Fig. 2, A and B). LPH⌬4 partially acquired endo H resistance, compatible with complex glycosylation of this deletion mutant in the Golgi apparatus. By contrast, the mutant lacking homologous domain III, LPH⌬3, was not transport-competent, and the introduction of a spacer containing seven glycines to avoid possible sterical hindrances did not alter its trafficking characteristics. Assessment of the proportions of the mannose-rich and complex glycosylated forms after scanning of the gels revealed a substantial increase in the proportion of the complex glycosylated LPH⌬4 as compared with the wild type counterpart (Fig. 2B). This was surprising because it indicated that the deletion of domain IV in LPH⌬4 leads to a more rapid processing of this deletion mutant than the wild type protein.
We further investigated the subcellular distribution of the mutant proteins in more detail by confocal laser microscopy. As shown in Fig. 2, LPH⌬3 was retained intracellularly and colocalized with the ER-DsRed marker (Fig. 2D). By contrast, and consistent with the biochemical data, LPH⌬4 colocalized with markers of the ER, the Golgi, and it was also detected at the plasma membrane, similar to the wild type protein (Fig. 2, C-E). The delivery of LPH⌬4 to the cell surface was assessed by probing its accessibility to trypsin in intact cells that were biosynthetically labeled for 4 h. Trypsin cleaves LPH at Arg 868 /Ala 869 to a protein band of 155-160 kDa (LPH␤ final ) ( Fig. 1A) (see Refs. 15 and 16), and for a review, see Ref. 20), which remains thereafter resistant to this protease. Because LPH⌬4 contains the trypsin site Arg 868 /Ala 869 , a cleaved product derived from this mutant should be generated provided that this protein is expressed at the cell surface. This cleavage product should be substantially smaller in size than its trypsin-cleaved wild type counterpart, LPH␤ final , due to the deletion of domain IV. Fig. 2F demonstrates that trypsin has cleaved LPH⌬4 in intact cells to a band of ϳ70 kDa. Similarly, wild type LPH was also cleaved at the cell surface to a band of ϳ155-160 kDa. In both cases, the wild type LPH and LPH⌬4, the mannose-rich forms were not cleaved, compatible with an intracellular localization of these protein forms and also with the specificity of the trypsin assay in cleaving only those forms that are exposed at the cell surface. Markedly, the majority of the complex forms of both wild type LPH and LPH⌬4 were cleaved by trypsin, implying a rapid transport of these proteins to the cell surface after terminal glycosylation.
Therefore, the differences in the kinetics of trafficking between wild type LPH and LPH⌬4 are restricted to the pathway between the ER and the Golgi. By contrast to wild type LPH and LPH⌬4 no cleavage products could be detected with LPH⌬3 (not shown), supporting the notion that this deletion form is located intracellularly.

Requirements for the LPH Deletion Mutants to Exit the ER
Dimerization of LPH in the ER is absolutely required for LPH to egress this organelle to the Golgi apparatus (35). The differential intracellular distribution and maturation patterns of the deletion mutants as well as the variable proportions of the glycoforms have altogether lead us to examine the quaternary structures of the mutants and assess their relevance to their transport out of the ER. Fig.  3A depicts the results obtained using sucrose density gradients. As has been previously shown, the mannose-rich LPH form was retained in the light as well as dense gradient fractions concomitant with its monomeric and dimeric states, respectively, and indicative of dimerization occurring along the early secretory pathway. The complex glycosylated protein, on the other hand, is detected exclusively in the dense fractions, indicating that the dimerization of the mannose-rich forms of LPH precedes its complex glycosylation and maturation in the Golgi (35). Surprisingly, the transport-competent LPH⌬4 deletion mutant did not require dimerization of its mannose-rich form in the ER prior to ER egress. As shown in Fig. 3A (the second top panel), the mannose-rich form of LPH⌬4 persisted as a monomeric protein, and the complex glycosylated LPH⌬4 initially appeared in the monomeric fractions. The majority of the complex glycosylated molecules were mainly found in the denser gradient fractions. Interestingly, complex glycosylated LPH⌬4 was revealed in two peaks in the gradient, compatible with two quaternary states, a dimeric and presumably a tetrameric state. A tetrameric LPH⌬4 form would be in line with the results obtained by Panzer et al. (36) for the LPH1646MACT mutant lacking 236 amino acids at the C terminus of homologous domain IV. By contrast, LPH⌬3 was exclusively detected in the lighter fractions of the gradients in its mannose-rich glycoform, compatible with retention in the ER as a monomeric protein.

Transport Kinetics of LPH and Deletion Mutants
We next analyzed the transport kinetics of the mutants in comparison with wild type LPH in pulse-chase experiments. Here, the immunoprecipitated proteins were treated with endo H to clearly discriminate between the mannose-rich and complex glycosylated forms. Complex glycosylated endo H-resistant LPH⌬4 appeared within 1.5 h of chase, and its proportion was substantially higher than its counterpart in the wild type protein (Fig. 3B, compare also Fig. 2, A and B), indicating that it is more efficiently transported to the Golgi apparatus than wild type LPH. By contrast, LPH⌬3 persisted as an endo H-sensitive mannose-rich polypeptide, compatible with ER localization. The Gly spacer containing LPH⌬3-7xGly mutant also revealed similar biosynthetic features as LPH⌬3 (not shown).

Folding of the Deletion Mutants
The variations in the quaternary structure of the deletion mutants as well as in their transport kinetics raised the question of causal folding variations. We therefore examined the folding of these mutants by using three procedures. In the first, the mutants were probed for their protease sensitivity using trypsin; in the second procedure, the enzymatic activities of the mutants were measured; and finally, in the third procedure, reactivity of the mutants with epitope-specific antibodies was assessed.
Trypsin Treatment-The tryptic digestion patterns of the wild type and mutant proteins are depicted in Fig. 3C. Wild type LPH was digested to two main bands corresponding to cleaved mannose-rich and complex glycosylated LPH (see also Ref. 3). This pattern did not change with prolonged digestion times. Similarly, LPH⌬4 pattern was also cleaved to two protein products that correspond to the mannose-rich and complex glycosylated forms. The smaller apparent molecular weights products fit well with a reduction corresponding to the size of the deleted domain IV. In a fashion similar to wild type LPH, the cleaved products of LPH⌬4 were also resistant to trypsin. Importantly, the cleavage of LPH⌬4 to the final products was not preceded by major intermediate cleaved forms, suggesting that one major trypsin site is exposed in the deletion mutant, which is in all likelihood the same as that in wild type LPH. By contrast to wild type LPH and LPH⌬4, LPH⌬3 was completely degraded by trypsin already after 1 min of treatment concomitant with the exposure of several trypsin cleavage sites and thus altered folding in comparison with wild type LPH and LPH⌬4.
Enzymatic Activities of LPH Deletion Mutants-Another approach to examine the folding and maturation pattern of a protein is to assess its biological function. We therefore analyzed the enzymatic activities of lactase and phlorizin hydrolase in these mutants in comparison with their wild type counterparts (Fig. 4). LPH⌬4 revealed slightly reduced activities of phlorizin hydrolase. The lactase activity was as expected absent because the lactase active site is found in residue Glu 1749 of domain IV. The lactase activity in LPH⌬3 was not detected. The data provide another support for malfolded LPH⌬3 and correct folding of LPH⌬4.
Epitope Mapping of Domain Deletion Mutants-The deletion mutants were immunoprecipitated with a panel of mAbs, which are specific in recognizing native or unfolded conformations of LPH (35). The control samples utilized immunoprecipitation of the GFP-tagged mutants with anti-GFP. Fig. 5 shows that LPH⌬4 and LPH⌬3 were isolated with anti-GFP FIGURE 3. Structural and functional features of LPH deletion mutants in COS-1 cells. A, assessment of the quaternary structure. Transiently transfected COS-1 cells were biosynthetically labeled and solubilized in 6 mM dodecyl-␤-m-maltoside. Cell lysates were layered on a sucrose density gradient. After centrifugation for 18 h at 100,000 ϫ g, fractions were collected, immunoprecipitated, and analyzed on SDS-PAGE. WT, wild type. B, transport kinetics of wild type LPH and mutant proteins. Transfected COS-1 cells were pulse-labeled for 1.5 h with [ 35 S]methionine and chased for the indicated periods of time with cold methionine. The immunoprecipitates were treated with endo H or not treated and analyzed by SDS-PAGE on 6% slab gels. C, trypsin sensitivity assay of wild type and LPH mutants. Transiently transfected COS-1 cells were biosynthetically labeled followed by immunoprecipitation of LPH proteins from the cell lysates. The immunoprecipitates were treated with trypsin for different times and analyzed by SDS-PAGE on 7% slab gels.

FIGURE 4. Enzymatic activity of deletion mutants (⌬-mutants) of LPH.
COS-1 cells were transiently transfected. 48 h after transfection, labeled cells were lysed, and proteins were immunoprecipitated. Immunoprecipitates were incubated with lactose and phlorizin, respectively, and the lactase and phlorizin hydrolase activities were measured by determining the concentration of released glucose by HPLC. The enzyme activities of the mutants were compared with those of wild type (WT) LPH. Error bar represents S.E.
antibody. Surprisingly, none of the mAbs against LPH recognized LPH⌬3, even the two mAbs, MLac6 and MLac10, that recognize unfolded and denatured forms of LPH. LPH⌬4, on the other hand, reacted with all the antibodies utilized with the exception of MLac6 and MLac10. Given that the antibodies were raised against the mature form of LPH, i.e. LPH␤ that comprises the two domains III and IV, it is obvious that all antibodies except MLac6 and MLac10 possess epitopes in domain III of LPH. Because MLac6 and MLac10 are directed against unfolded forms of LPH, the results indicate that LPH⌬4 is properly folded, lending a strong support to the protease sensitivity data. LPH⌬3, on the other hand, is malfolded and is therefore not recognized by the antibodies. It is also likely that none of epitopes is found on LPH⌬3. This view is supported by the observation that LPH⌬3 does not react with MLac6 or MLac10, which are directed against malfolded forms of LPH.

Domain III Is a Transport-competent and Functional Protein
The data gathered so far strongly suggest that domain III is a central autonomous component of LPH. The next step was therefore to express this domain independently and examine its trafficking and functional properties. As shown in Fig. 6B, domain III expression in COS-1 cells revealed a predominant endo H-and endo F-sensitive protein band in the cell lysates, indicating that it is a mannose-rich glycosylated form of domain III. The cell culture medium contained an endo H-resistant and endo F-sensitive protein, compatible with a complex glycosylated domain III. These results clearly indicate that domain III is secreted into the cell exterior immediately and rapidly upon maturation in the Golgi apparatus. To substantiate the data with a further approach, immunofluorescence images were generated. Fig. 6C shows that domain III was located in the ER, compatible with the major mannose-rich form in the cell lysates. When the cells were subjected to a 20°C temperature block, domain III was found in the Golgi apparatus. Assessment of the quaternary structure of domain III, performed at 20°C to analyze mannose-rich and complex glycosylated proteins, revealed monomeric forms of the mannoserich protein as well as the complex glycosylated form (Fig. 7A). It should be noted that the overall labeling intensity of the complex glycosylated protein in all the lanes as compared with the mannose-rich polypeptide did not comprise more than 10% of total domain III in the cell lysates.
Continuous metabolic labeling was performed to determine the transport rate of domain III. As shown in Fig. 7B, domain III appeared in the medium after 90 min of labeling. Finally, we probed the folding of domain III utilizing trypsin sensitivity and measurement of its enzymatic activity. Fig. 7C shows that domain III is predominantly resistant to trypsin. Its phlorizin hydrolase activity is, however, reduced by about 50% (Fig. 8).

Sorting of Domain III in Polarized MDCK Cells
LPH is sorted into the apical membrane in polarized MDCK cells and intestinal cells with high fidelity. Because it is proposed that putative apical sorting signals are located in the ectodomain of the LPH mature form and domain III builds one-half of LPH␤, we analyzed its sorting in a polarized cell line to determine whether or not this region contains putative signals for apical sorting of LPH. Domain III was stably expressed in MDCK cells, and its sorting was analyzed in a membrane filter system as described previously (38) . Fig. 9, A and B, demonstrate that domain III is secreted predominantly at the apical surface of MDCK cells. In fact, more than 80% of this protein was found at the apical side, indicating that the sorting of domain III is not as efficient as for wild type LPH. These data suggest that domain IV is most likely devoid of putative apical sorting signals.

DISCUSSION
Characterization of the structure, biosynthesis, and trafficking of the individual subdomains within LPH, an essential brush-border membrane enzyme, constitutes an important step toward understanding its function and impact to the intestinal epithelial cell physiology. Given that the three-dimensional structure of LPH has not been elucidated yet, alternative approaches have to be designed to determine the significance and relevance of the individual subunits to each other in the context of this multiple domain protein.
Lactases, or more properly, ␤-galactosidases, are grouped within four of the nearly 100 families of glycosyl hydrolases (GHs) that have been characterized (39). Mammalian intestinal lactase (LPH) is classified in the GH1 family, along with enzymes present in a variety of organisms acting against different types of ␤-glycosides. Although most GH1 enzymes (mostly bacterial) thus far characterized are conformed by a single domain, intestinal lactase is synthesized as a multidomain precursor protein that is encoded by a gene resulting from the fusion of four tandemly arranged repetitions of an ancestor gene (1,32). Maturation of the enzyme generates the brushborder form comprised by the last two domains, III and IV (2), each with differential activity. Domain III shows specificity toward glycosides, such as phlorizin, whereas domain IV is specifically active against lactose (22). Interestingly, individual expression of domain III, but not domain IV (data not shown), of LPH reveals a correctly folded, transport-competent, and rapidly secreted molecule, thus underscoring its autonomous character that has been conserved from prokaryotes to eukaryotes.
Despite its strong homologies with domain III, domain IV is not a folding-competent, a transport-competent, or an enzymatically active species per se. This domain, however, plays a central regulatory role in the context of the function and trafficking of LPH. It contains the LAC236 stretch that is required for dimerization of LPH (24,36). The essential function of domain IV within the LPH complex becomes evident when considering its role in the dimerization of LPH. In fact, domain III does not dimerize, and the phlorizin hydrolase activity of this domain is elevated by a factor of 2.5-fold in the dimeric LPH molecule. Additionally, domain IV is rate-limiting along the FIGURE 7. Structural and functional features of D3. A, assessment of the quaternary structure of D3. Transiently transfected COS-1 cells were biosynthetically labeled at 20°C to avoid secretion and solubilized in 6 mM dodecyl-␤-m-maltoside. Cell lysates were layered on a sucrose density gradient. After centrifugation for 18 h at 100,000 ϫ g, fractions were collected, immunoprecipitated, and analyzed on SDS-PAGE. B, transport kinetics of D3. Biosynthetically labeled proteins were immunoprecipitated from cell lysates and cell culture media after the indicated labeling times and analyzed by SDS-PAGE. lys, lysate; med, medium. C, trypsin sensitivity assay with D3. Transiently transfected COS-1 cells were biosynthetically labeled followed by immunoprecipitation of LPH proteins from cell lysates and cell culture media. The immunoprecipitates were treated with trypsin for different times and analyzed by SDS-PAGE. The immunoprecipitates were incubated with phlorizin, and the phlorizin hydrolase activity was measured by determining the concentration of the released glucose by HPLC. The enzyme activity of the mutant was compared with wild type LPH. Error bar represents S.E.; n ϭ 4. Proteins that have been secreted into the apical (ap) and the basolateral (bl) medium, respectively, were isolated by adding anti-LPH mAb and protein A-Sepharose to the collected media. Intracellular (ic) proteins were isolated by immunoprecipitation after cell lysis. The immunoprecipitates were analyzed by SDS-PAGE on 9% slab gels followed by phosphorimaging. B, the quantification of secreted D3 was performed with the Quantity Oneா software from Bio-Rad. The error bars represent S.E.; n ϭ 5. secretory pathway of LPH from the ER to the Golgi. In fact, LPH⌬4, a deletion mutant that lacks the entire homologous domain IV, acquires more rapidly complex glycosylation than its wild type counterpart, proposing a role of this domain in decelerating LPH processing. Importantly, LPH⌬4 does not dimerize in the ER, lending a strong support to the view that dimerization is initiated by homologous domain IV and supporting previous data that assigned LAC236 an essential role in dimerization. It is very likely therefore that the retarded trafficking of wild type LPH in comparison with LPH⌬4 is due to its dimerization prior to the ER exit, whereas this additional step is not required for LPH⌬4 to acquire transport competence. Interestingly, epitope mapping with a panel of mAbs against the mature brush-border form of LPH (domains III and IV) demonstrated a similar pattern of recognition for LPH⌬4 and LPH, strongly suggesting that the epitopes tested are located in domain III. Given that these antibodies react only with native LPH species, but not with denatured LPH on Western blots, our data strongly suggest that the tertiary structure of LPH⌬4 is comparable with its counterparts in human wild type LPH and that domain III represents the structural core of the mature protein.
On the other hand, domain IV is not an autonomous region. It harbors the lactase catalytic site at Glu 1749 (9,22) and acquires activity only when LPH dimerizes (35). Therefore, domain IV plays a role as a regulatory switch that triggers the dimerization of the LPH molecule, thus activating itself and elevating the phlorizin hydrolase activities in domain III.
Initial cleavage of mature LPH in the Golgi occurs between Arg 734 and Leu 735 and generates LPH␤initial that is transported with high fidelity to the apical membrane. In the intestinal lumen, LPH␤initial undergoes another cleavage at Arg 868 / Ala 869 by pancreatic trypsin to generate LPH␤final (15,16). The significance of the stretch between Leu 735 and Arg 868 in the context of trafficking and sorting of LPH has been until present obscure. Our data assign a role to this region in association with domain III in the sorting of LPH to the apical membrane. In this respect, it is interesting to note that domain III per se is not as efficiently transported to the apical membrane in polarized MDCK cells as wild type LPH, strongly proposing that the polypeptide stretch Leu 735 -Arg 868 in domain II could be important for the fine-tuning of polarized sorting.
Our data present a hierarchical model of LPH in which the homologous domain III constitutes a fully autonomous core domain within the LPH molecule. In addition, it represents another intramolecular chaperone of LPH besides the profragment LPH␣ (Fig. 10). This model assumes that the profragment (2) and homologous domain III (this study) attain their native conformation autonomously. Although domain III is transport-competent and enzymatically active per se, it requires homologous domain IV for elevation of its enzymatic activity and regulation of its trafficking kinetics. This occurs via dimerization of the entire LPH molecule, an event that is triggered by domain IV. Nevertheless, domain IV is a non-autonomous domain that cannot fold independently; it requires the profragment as well as domain III as templates for correct folding. Correctly folded domain IV is now capable of triggering the dimerization of LPH in the ER (35), an event that is required for LPH to exit the ER, for regulation of its transport kinetics and elevation of its enzymatic activities. The profragment and domain III function therefore act as switches for correct folding of domain IV, which in turn "pays back" by giving domain III an increased phlorizin hydrolase activity and gaining more activity as the lactase active site. To our knowledge, this is the first example of a mechanism in which a protein has two intramolecular chaperones and is not activated by propeptide cleavage, as described for zymogens, neuropeptides, and prohormones (33,37), but by intramolecular organization and oligomerization.