The SANT2 domain of the murine tumor cell DnaJ-like protein 1 human homologue interacts with alpha1-antichymotrypsin and kinetically interferes with its serpin inhibitory activity.

The murine tumor cell DnaJ-like protein 1 or MTJ1/ERdj1 is a membrane J-domain protein enriched in microsomal and nuclear fractions. We previously showed that its lumenal J-domain stimulates the ATPase activity of the molecular chaperone BiP/GRP78 (Chevalier, M., Rhee, H., Elguindi, E. C., and Blond, S. Y. (2000) J. Biol. Chem. 275, 19620-19627). MTJ1/ERdj1 also contains a large carboxyl-terminal cytosolic extension composed of two tryptophan-mediated repeats or SANT domains for which the function(s) is unknown. Here we describe the cloning of the human homologue HTJ1 and its interaction with alpha(1)-antichymotrypsin (ACT), a member of the serine proteinase inhibitor (serpin) family. The interaction was initially identified in a two-hybrid screening and further confirmed in vitro by dot blots, native electrophoresis, and fluorescence studies. The second SANT domain of HTJ1 (SANT2) was found to be sufficient for binding to ACT, both in yeast and in vitro. Single tryptophan-alanine substitutions at two strictly conserved residues significantly (Trp-497) or totally (Trp-520) abolished the interaction with ACT. SANT2 binds to human ACT with an intrinsic affinity equal to 0.5 nm. Preincubation of ACT with nearly stoichiometric concentrations of SANT2 wild-type but not SANT2: W520A results in an apparent loss of ACT inhibitory activity toward chymotrypsin. Kinetic analysis indicates that the formation of the covalent inhibitory complex ACT-chymotrypsin is significantly delayed in the presence of SANT2 with no change on the catalytic efficiency of the enzyme. This work demonstrates for the first time that the SANT2 domain of MTJ1/HTJ1/ERdj1 mediates stable and high affinity protein-protein interactions.

The rough endoplasmic reticulum (ER) 1 is the primary site for the synthesis and maturation of secreted and membrane proteins. At this site molecular chaperones and their associated enzymes promote the folding and assembly of newly synthesized polypeptides (1,2). Only native proteins leave the ER to enter the vesicular secretory pathway and reach their appropriate destinations. Misfolded or aberrant proteins are transported back to the cytosol, then transferred to the 26 S proteasome in a process termed ER-associated protein degradation (3)(4)(5)(6).
The molecular chaperone immunoglobulin heavy chain-binding protein (BiP)/GRP78, a member of the Hsp70 family resident of the ER, is involved in many cellular processes that include the translocation of newly synthesized polypeptides across the ER membrane, participation in their folding and maturation, assisting in refolding and renaturation, targeting misfolded proteins to the cytosol for proteasomal degradation, maintaining selectivity of the ER membrane by closing the translocon pore, as well as regulating calcium homeostasis (7)(8)(9)(10). ATPase activity of BiP is required for most of these processes (11,12). In the ATP-bound form BiP binds to unfolded substrates with low affinity. The hydrolysis of ATP to ADP induces conformational changes that stabilize the BiPunfolded substrates complexes. The ATPase activity of BiP and other Hsp70 proteins is stimulated by members of the J-domain family (13). In the yeast Saccharomyces cerevisiae BiP/ Kar2p is assisted by at least three J-domain proteins: Sec63p, Jem1p, and Scj1p (12, 14 -17). The integral membrane protein Sec63p is an essential component of the ER membrane translocon (18). Jem1p together with BiP/Kar2p are part of a nuclear fusion complex (19,20), and Scj1p is likely to be involved in protein folding and assembly in the ER lumen (17,21). Both Jem1p and Scj1p may also facilitate the retrotranslocation of lumenal ER-associated protein degradation substrates to the cytosol by preventing aggregation of misfolded polypeptides in the ER (12). In mammals, five ER J-domain proteins have been identified so far, all of which stimulate BiP ATPase activity: MTJ1/ERdj1 (22)(23)(24), the human Sec63 homologue/ERdj2 (25,26), the Scj1 homologue or HEDJ/ERdj3 (21,27), ERdj4 (28), and ERdj5 (29).
We previously showed that the lumenal J-domain of murine MTJ1 forms a stable complex with BiP and stimulates its ATPase activity at stoichiometric concentrations (23). More recently, Zimmermann and colleagues (24) identified MTJ1 as having higher affinity for BiP than the more abundant Sec63 homologue in dog pancreas. The NH 2 -terminal of MTJ1 that carries the J-domain is followed by one transmembrane helix and a cytosolic carboxyl-terminal domain composed of a spacer region that possesses homology with Sec63p flanked by two tryptophan-mediated repeats or SANT domains for which no function has been attributed (23,24). SANT domains are struc-turally related to the Myb DNA-binding domain (30,31) and are present at one to five copies in a variety of transcription factors or regulators including Swi3, Ada2, N-CoR, TFIIIB (30,(32)(33)(34) or components involved in chromatin remodeling (35,36). Besides these reports, very little is known about the function of SANT-containing proteins.
In the present study, we have cloned the human homologue of MTJ1, which we call HTJ1, and used its COOH-terminal domain as bait in a two-hybrid screening of a human liver cDNA library. We have identified ␣ 1 -antichymotrypsin (ACT), a member of the serpin family, as a potential interacting candidate. Targeted mutagenesis indicates that the second SANT domain of HTJ1 (SANT2) mediates the interaction with ACT and that this interaction can be totally disrupted by a tryptophan to alanine single mutation (W520A). SANT2-ACT interaction was confirmed in vitro using purified proteins. ACT once bound to SANT2 was found to have no inhibitory activity toward chymotrypsin. SANT2 significantly slows down the kinetic of formation of the ACT-chymotrypsin acyl complex with no significant effect on the catalytic efficiency of the proteinase. This report documents for the first time that the human homologue of MTJ1/ERdj1 can mediate high affinity protein-protein interactions through its cytosolic domain. In another report, we describe that HTJ1 also interacts with a member of the inter-␣-trypsin inhibitor and protects it from being processed by its natural protease. 2 Our data suggests that HTJ1/ERdj1 may play a role in the secretion of protease inhibitors in mammalian cells.

EXPERIMENTAL PROCEDURES
Materials, Antibodies, Proteins, and Strains-All components of the two-hybrid system, including cloning vectors pGBKT7 and pACT2, human liver cDNA library, S. cerevisiae strain AH109, anti-c-Myc monoclonal antibody, anti-HA tag polyclonal antibody, X-␣-gal, dropout supplements, and cobalt-chelating resin Talon were obtained from Clontech. Bacterial vector pUC19 was from New England BioLabs, Beverly, MA. pET15b plasmid and the BL21(DE3) codon ϩ strain were from Novagen, Madison, WI. Pfu polymerase was from Stratagene, La Jolla, CA. All restriction enzymes and T4 DNA ligase were from MBI Fermentas, Amherst, NY. The ECL kit was from Amersham Biosciences. Human plasma ACT was from Calbiochem-Novabiochem Corp., San Diego, CA. Bovine pancreas chymotrypsin, and its substrate succinyl-Ala-Ala-Pro-Phe-p-nitroanilide (Succ-AAPF-pNA) were from Sigma. Monoclonal anti-␣ 1 -antichymotrypsin antibody was obtained from Fitzgerald Industries International, Inc. Concord, MA.
Cloning of HTJ1 cDNA-A nucleotidic sequence from a 10-week-old human embryo (AK027263, GI: 14041830) coding for a putative MTJ1 homologue was used as a template to design the set of primers for the amplification of a human liver cDNA library. The forward primer (5Ј-ATGACGGCTCCTTGCTCCCAG-3Ј) and reverse primer (5Ј-TCAGCT-TTTAGCTTGTTTTTTCTTTTGGACC-3Ј) were used in PCR with Pfu polymerase. The PCR product (1700 bp) was then subcloned into the SmaI site of pUC19. The construct (pUCHTJ1) was sequenced on both strands and sequence analysis was carried out using BLAST sequence analysis software.
Yeast Two-hybrid Library Screening-The construct pGBKT7/ HTJ1 242-554 (TRP1) was used as bait to screen a human liver cDNA library fused to the pACT2 (LEU2) GAL4-activation domain and the HA epitope tag. The bait pGBKT7/HTJ1 242-554 and the library-containing target pACT2 vector were sequentially transformed into the two-hybrid yeast strain AH109 as described (38). As strain AH109 possesses three reporter genes (ADE2, HIS3, and MEL1) that are under the control of the Gal4 upstream activating sequences and TATA boxes, transformants were grown on minimal synthetic dropout four medium lacking tryptophan (Trp), leucine (Leu), histidine (His), and adenine (Ade) (SD/ϪTrp/ϪLeu/ϪHis/ϪAde) to select for potential interactors. Approximately 2 ϫ 10 6 clones, representing 1/3 library equivalents, were screened against the pGBKT7/HTJ1 242-554 bait. Transformants with the ability to grow on SD/ϪTrp/ϪLeu/ϪHis/ϪAde that turned blue in the presence of X-␣-gal were analyzed further as described under "Results." Western Blot Analysis of Protein Expression in Yeast Protein Extracts-The yeast strain AH109 was transformed separately with pGBKT7 or pACT2-derived constructs (Table I) and grown overnight on selective SD medium at 30°C. The yeast protein extracts were prepared according to the manufacturer's procedure (Clontech, protocol number PT3024-1, version PR91200). Following electrophoresis on a 10% SDS-PAGE gel, proteins were transferred onto nitrocellulose membrane in 50 mM Tris (pH 8.0), 380 mM glycine, 0.1% SDS, and 20% methanol (transfer buffer). The recombinant proteins expressed from pGBKT7derived constructs were detected using an anti-c-Myc monoclonal primary antibody and an anti-mouse conjugated to horseradish peroxidase secondary antibody. The expression of the pACT2-derived fusion proteins was detected by using a rabbit polyclonal antibody to HA. Immunoreactivity was visualized by enhanced chemiluminescence using the ECL kit.
Expression and Purification of Recombinant Proteins-The recombinant histidine-tagged proteins were expressed from pET15-derived constructs (Table I). The E. coli BL21(DE3) codon ϩ cells were grown at 37°C in Luria broth (LB) supplemented with carbenicillin at 100 g/ml final concentration. Synthesis of recombinant proteins was induced by addition of isopropyl-1-thio-␤-D-galactopyranoside at 0.4 mM final concentration at mig-log exponential phase (A 600 ϳ 0.5-0.6). Cells were harvested after a 16-h induction, lysed using a French press at 10,000 p.s.i., and the histidine-tagged protein was purified on cobalt-chelating resin as described by the manufacturer (Talon resin, Clontech). All recombinant proteins were expressed in the soluble fraction except for His 6 -ACT 1-400 , which accumulates in inclusion bodies and was purified from lysate pellets resolubilized in 20 mM Tris (pH 8.0), 8 M urea, 100 mM NaCl and purified in denaturing conditions. Fractions eluted from the resin were dialyzed overnight at 4°C in 20 mM Tris (pH 8.0), 100 mM NaCl to promote renaturation. The renatured soluble recombinant ACT was further purified by fast pressure liquid chromatography using gel filtration on Superdex 75 (Amersham Biosciences) equilibrated in the same buffer. Protein concentrations were estimated by the method of Bradford (39). Protein purity and homogeneity were analyzed by PAGE performed in both denaturing and native conditions (40).
Dot Blot Protein Analysis-Dot blots were performed as described (41) with the following modifications. Various amounts of recombinant His 6 -HTJ1 proteins were applied onto a nitrocellulose membrane, dried, then incubated overnight with a solution of soluble renatured His 6 -ACT 1-400 (5 g/ml) in 20 mM Tris-Cl (pH 8.0), 150 mM NaCl. After washing in phosphate-buffered saline, bound ACT was detected with an anti-human ACT monoclonal antibody (1:20,000). In control experiments, purified His 6 -J-MTJ1 and His 6 -J-MTJ1:H89Q prepared as described (23) were applied to the nitrocellulose membrane, and incubated with a solution of recombinant His 6 -BiP (5 g/ml) in Tris-Cl (pH 8.0), 150 mM NaCl. Bound BiP was detected by chemiluminescence using a commercially available anti-BiP antibody (PAI-014, Affinity Bioreagents Inc., Golden, CO). In all experiments, immunoreactivity was visualized by ECL.
Native PAGE and Immunodetection of ACT-HTJ1 Interaction-Human plasma ACT (0.64 M) was incubated at 25°C with increasing concentrations of purified His 6 -HTJ1 493-554 wild-type or mutant (His 6 -HTJ1 493-554 :W497A) in 20 mM Tris (pH 8.0), 100 mM NaCl prior to electrophoresis in non-denaturing conditions (40). ACT was detected by Western blotting using a commercially available monoclonal antibody.
Tryptophan Fluorescence Quenching Measurements-All measurements were carried out on a Cary Eclipse fluorescence spectrophotometer at a temperature of 25°C regulated by a thermostatted cell holder and a Cary PCB15 Water peltier system (Varian Instruments, Walnut Creek, CA). All data were analyzed using the Cary Eclipse software (Varian Instruments). The emission spectra were recorded between 300 and 450 nm at an excitation wavelength of 280 nm. The excitation and emission slits were set to 10 nm. The emission wavelength increment was 4 nm and the acquisition time 3 s. Tryptophan fluorescence intensities were corrected for the contribution of ACT alone analyzed in the same conditions of temperature, buffer (20 mM Tris (pH 8.0), 100 mM NaCl), and concentrations. In titration experiments, the fraction of ACT bound to SANT2 (4 nM) at every serpin concentration (0 -16 nM) was calculated relative to the fluorescence intensity (excitation wavelength ϭ 280 nm; emission wavelength ϭ 350 nM) at saturating concentrations of ACT (100% bound) minus the intensity of SANT2 alone (0% bound). The intrinsic dissociation equilibrium constant, K diss , was determined using the Scatchard representation corresponding to the equation, gives a straight line with a negative slope that equals Ϫ1/K diss [ACT]b and extrapolation on the x axis gives the maximum concentration of bound ACT and is used to deduce the stoichiometry of the complex. As SANT2 and ACT contain two and three tryptophans, respectively, we used Lehrer representation to determine the fraction of tryptophan residues quenched upon SANT2-ACT interactions as described (42), using the equation, where F 0 is the tryptophan fluorescence of SANT2 alone; F is the tryptophan fluorescence of the SANT2-ACT complex at a given ACT concentration [ACT]; ƒ a is the maximum fraction of accessible fluorophore; K is the Lehrer quenching constant of an individual tryptophan.
gives a straight line whose extrapolation on the y axis ([ACT] ϭ 0) corresponds to the value of 1/ƒ a (42). Chymotrypsin Activity-All steady state experiments were carried out at 25°C in 20 mM Tris (pH 8.0), 100 mM NaCl with the chromogenic substrate Succ-AAPF-pNA. Absorbance was recorded spectrophotometrically at 405 nm. K m and V max were determined using the Lineweaver-Burk representation and the Michaelis-Menten equation at 4 nM chymotrypsin, substrate ranged from 0 to 0.1 mM, in the absence or presence of 4 nM SANT2. The catalytic constant, k cat , was determined for 0.2-5.0 nM chymotrypsin in the presence of 0.1 mM Succ-AAPF-pNA. To determine the residual activity of chymotrypsin in the presence of the preformed ACT-SANT2 complex, human plasma ACT (4.0 nM) was preincubated for 20 min at 25°C in the presence of 0 -640 nM His 6 -HTJ1 493-554 , His 6 -HTJ1 493-554 :W497A, or His 6 -HTJ1 493-554 :W497A, W520A and 0.1 mM Succ-AAPF-pNA prior to the addition of chymotrypsin (4.0 nM final concentration). The residual chymotrypsin activity was measured for 5-10 min at 15-s intervals. The percentage of inhibition was normalized to the activity of chymotrypsin alone (100%). Kinetic analysis of ACT inhibition of chymotrypsin in the presence or absence of SANT2 was performed as described (43) with the following modifications. Equimolar concentrations of chymotrypsin and ACT (4 nM) were incubated 0 -10 min and the residual chymotrypsin activity was measured at timed intervals after addition of 0.1 mM Succ-AAPF-pNA. The concentration of free chymotrypsin [E f ] was estimated using a standard curve of chymotrypsin activity. Using the equation,

Isolation, Sequence Analysis of cDNA Encoding the HTJ1, and Amino Acid Sequence Similarities of HTJ1 and MTJ1-
The human homologue of MTJ1/ERdj1 was cloned from a human liver cDNA library (see "Experimental Procedures") (AY225122) and found to be 84% identical to the murine protein (Fig. 1). An NH 2 -terminal signal sequence containing a putative cleavage site between amino acids 47 and 48 could target the signature J-domain (residues 63-138) to the lumen of the endoplasmic reticulum where it can interact with BiP (22)(23)(24). The J-domain is followed by a membrane anchor, a cytosolic NH 2 -terminal region that may interact with ribosomes (residues 173-239) as observed in the murine homologue MTJ1 (24), and a large cytosolic domain consisting of three additional subdomains or regions, namely two SANT domains (SANT1 residues 325-377 and SANT2 residues 493-545), also called tryptophan-mediated repeats, separated by a 116-residue long spacer that includes a region (residues 409 -465) that exhibits 28% identity with Sec63p. A KKXX-like tail motif would account for HTJ1 retrieval from the Golgi to the ER and its retention, as observed for other type 1 membrane proteins (44 -46).
Interaction of HTJ1 with ACT Detected by the Two-hybrid System in Yeast-The liver is an organ important in the secretion of many plasma proteins and molecular chaperones involved in these processes, such as BiP and possibly HTJ1, are particularly abundant in liver microsomes. To elucidate the function of HTJ1 in secreting cells, we used a yeast two-hybrid system and screened a human liver cDNA library for proteins that interact with the carboxyl-terminal HTJ1 242-554 fragment ( Fig. 2A). Prior to screening, the yeast strain AH109 transformed with the vector pGBKT7/HTJ1 242-554 (Table I) and the empty pACT2 vector was shown to be transcriptionally inert by nutrition selection (data not shown). The HTJ1 242-554 fragment was efficiently expressed in-frame with the Gal4 DNA-binding domain and the c-Myc epitope, as indicated by the size of the fusion protein and its immunoreactivity toward an anti-c-Myc antibody (Fig. 2B). Next, the pGBKT7/HTJ1 242-554 was used as bait for the screening of a human liver cDNA library cloned in the pACT2 plasmid. Initially, 23 yeast clones induced the expression of the three reporter genes (HIS3, ADE2, and MEL1), indicated by growth on SD dropout four medium and blue staining in the presence of X-␣-gal. Of these, 15 were eliminated from further studies because the isolated plasmids retransformed in the AH109 strain activated the reporter genes in the absence of the bait. Putative interacting targets were identified by DNA sequencing. Four clones corresponded to metallothionein (BC008408), a protein that has been isolated in numerous screens and is thus involved in multiple, nonspecific interactions (41). 3 One clone designated pACT2/ ACT 140 -400 (Fig. 3A) contained a large insert coding for residues 140 -400 of the ACT (AF089747) (47) in the same reading frame as the NH 2 -terminal Gal4 activation domain and the HA epitope tag of the pACT2 vector (Fig. 3B). The yeast strain AH109 co-transformed with pACT2/ACT 140 -400 and the vector pGBKT7/53 did not grow on the selective medium, SD/ϪTrp/ ϪLeu/ϪHis/ϪAde/X-␣-gal (Fig. 4, last lane), indicating that clone pACT2/ACT 140 -400 could not activate the reporter gene system on its own and that ACT may represent a true HTJ1interacting protein.
ACT Interacts with the SANT2 Domain of HTJ1-The region of HTJ1 responsible for interaction with ACT was further characterized in vivo and a number of HTJ1 variants were engineered including a SANT2-less (pGBKT7/HTJ1 242-493 ), a SANT2-only (pGBKT7/HTJ1 493-554 ), and SANT1-only fusion proteins (pGBKT7/HTJ1 242-411 )( Fig. 2A and Table I). Efficient expression of each fusion protein in yeast extracts was detected by immunoblot using the c-Myc antibody (Fig. 2B). Yeast strain AH109 co-transformed with pACT2 140 -400 and all pGBKT7derived constructs grows on medium lacking leucine and tryptophan (Fig. 4), indicating that the two plasmids are efficiently propagated in those conditions. However, the strain can only grow on SD/ϪTrp/ϪLeu/ϪHis/ϪAde medium and form blue colonies in the presence of X-␣-gal with constructs that express the SANT2 domain (residues 493-554) (Fig. 4). Control plasmids expressing the lumenal J-domain of HTJ1 (pGBKT7/J-HTJ1  ) or the p53 antigen (pGBKT7/53) do not sustain growth on selective medium when co-expressed with the ACT 140 -400 fragment (Fig. 4). Additionally, the single substitution of the highly conserved tryptophan residue at position 520 by alanine within the SANT2 domain totally abolished the transcription of the reporter genes (pGBKT7/HTJ1 493-554 : W520A, Fig. 4), and the same substitution at position 497 greatly destabilizes the complex because only marginal growth was observed (pGBKT7/HTJ1 493-554 :W497A, Fig. 4). As expected, the double mutant is transcriptionally inactive (pG-BKT7/HTJ1 493-554 :W497A,W510A, Fig. 4). Expression of the reactive site loop (RSL) (pACT2/ACT 344 -400 ) is not sufficient for growth on selective medium and its deletion (pACT2/ ACT   , Fig. 4) does not abolish interaction with SANT2. These results indicate that the SANT2 domain of HTJ1 interacts with a region in ACT comprised between residues 140 and 343 that does not overlap with the RSL and does not require structural integrity of the serpin.
We then analyzed the interaction of SANT2 with human plasma-glycosylated ACT, using native electrophoresis and Western blot (Fig. 7). A SANT2-ACT complex is readily detectable as a low migrating species on native polyacrylamide gels (Fig. 7A, lane 2), and all ACT is bound in the presence of five molar excess or more of SANT2 (Fig. 7A, lanes 4 -8). A complex also forms in the presence of SANT2:W497A (Fig. 7B), but less efficiently as not all ACT was trapped in the presence of 80 molar excess of SANT2 mutant (Fig. 7B, lane 8). In these conditions, no complex was observed for the double mutant   Interaction of SANT2 with ACT is accompanied by a decrease in SANT2 tryptophan fluorescence (Fig. 8A). The fluorescence signal declines with increasing concentrations of ACT until a plateau is reached at saturating concentrations of serpin (Fig. 8B). Using Scatchard representation and equations (see "Experimental Procedures"), we determined that the SANT2-ACT complex forms with an intrinsic affinity equal to 0.5 nM and a serpin-SANT2 stoichiometry of 1:1 (Fig. 8C). SANT2 and ACT contain two and three tryptophan residues, respectively. Using the Lehrer representation (Fig. 8D) and the equations described under "Experimental Procedures," we calculated that the fraction of accessible tryptophan f a equals 0.2, indicating that only one tryptophan residue of the five present contributes to the change in fluorescence observed upon ACT-SANT2 interaction. Similar experiments repeated with the SANT2:W520A mutant did not give any fluctuation in signal upon ACT addition. Taken together these data indicate that SANT2 binds to ACT at one site, with an affinity in the subnanomolar range and that Trp-520 is crucial to mediate this interaction and contribute to most of the fluorescence quenching signal.
SANT2 Alters ACT Inhibitory Activity-We next investigated whether the HTJ1 493-554 fragment interferes with the ACT inhibitory activity toward chymotrypsin. Indeed, equimolar concentrations of ACT totally inhibits chymotrypsin protease activity (Fig. 9A), whereas 160 molar excess of His 6 -HTJ1 493-554 (SANT2) wild-type or mutant had no effect on the serine protease activity (data not shown). Preincubation of ACT with increasing concentrations of the SANT2 fragment prevents ACT from exerting its inhibitory activity toward chymotrypsin (Fig. 9). Half-maximal chymotrypsin activity was restored at an ACT:SANT2 molar ratio equal to about 1:10. In similar conditions, the mutant His 6 -HTJ1 493-554 :W497A restores about 50% of the activity, whereas the double mutant His 6 -HTJ1493 -554 :W497A,W520A has no effect, even when present in a very large molar excess over ACT (Fig. 9B). SANT2 does not alter the catalytic properties of chymotrypsin and both K m and k cat remained unchanged (Table II).
SANT2 Delays the Formation of the ACT-Chymotrypsin Complex-Inhibition of the serine proteinase activity by serpin occurs with formation of an irreversible proteinase-serpin complex. This structure is stabilized by insertion of the cleaved NH 2 -terminal of the reactive site loop into ␤-sheet A (48). We examined the kinetics of formation of the ACT-chymotrypsin complex in the presence of millimolar concentrations of the chromogenic substrate. Incubation of equimolar concentrations of ACT and chymotrypsin (0.7 M) leads to a SDS-resistant complex that quickly accumulates within the 5-s to 1-min time range (Fig. 10A). A band slightly faster than free ACT is also noticeable after 1 min and most likely corresponds to cleaved ACT. When ACT was preincubated with 20 molar excess SANT2, the formation of the ACT-chymotrypsin complex is considerably delayed, accumulating only after 1 min and remaining incomplete after 10 min of incubation (Fig. 10B). Higher concentrations of SANT2 lead to more pronounced de- lays and barely any ACT-chymotrypsin complex could be detected in the presence of 80 molar excess of SANT2 over ACT (data not shown). The same experiments performed with the SANT2:W497A variant indicate that the mutant also delays the formation of the inhibitory ACT-chymotrypsin complex, although less efficiently than the wild-type SANT2 domain as some ACT-chymotrypsin complex appears after 20 s of incubation (Fig. 10C). In all experiments the amount of SANT2 remains unchanged, indicating that wild-type and mutant SANT2 are not substrates for chymotrypsin in these conditions and do not compete with ACT or the chromogenic substrates for binding to the proteinase. Finally, the rate constant of association between ACT and chymotrypsin is about two times FIG. 4. Yeast two-hybrid one-on-one interactions. HTJ1 constructs fused to the Gal4 DNA-binding domain (described in Fig. 2) were individually analyzed for interaction with fragments of ACT expressed in yeast as a fusion with the Gal4 activation domain (see Fig. 3). The prey plasmids pACT2/ACT 140 -400 , pACT2/ACT 344 -400 , or pACT2/ACT  (LEU2) were co-transformed into the auxotrophic S. cerevisiae AH109 strain with bait plasmid containing HTJ1 truncated constructs (TRYP1). Double transformants were selected on selective medium depleted in tryptophan and leucine (ϪTrp/ϪLeu). Interaction between the bait (HTJ1) and the prey (ACT) brings the Gal4 activation domain (AD) and the DNA-binding domain (DNA/BD) into close proximity, resulting in the activation of the HIS3, ADE2, and MEL1 reporter genes. Yeast cells harboring plasmids that code for interacting prey-bait pairs can grow on selective medium depleted in adenine and histidine (ϪTrp/ϪLeu/ϪAde/ϪHis) and turn blue in the presence of X-␣-gal because of the expression of the MEL1 gene product (␣-galactosidase). slower in the presence of SANT2 (Fig. 11, Table II). These results are consistent with our observations that incubation of human plasma ACT with SANT2 results in an apparent loss of ACT inhibitory activity because of a delay in the formation of the ACT-chymotrypsin suicide inhibitory complex. The mutant SANT2:W497A has reduced affinity for ACT, allowing more rapid release of the free serpin that can readily inactivate the proteinase in our assay conditions. DISCUSSION In mammals, BiP is a molecular chaperone resident of the ER that can potentially interact with at least five J-domain accessory proteins recently renamed ERdj1-5 (21)(22)(23)(24)(25)(26)(27)(28)(29). MTJ1/ ERdj1 is a class III membrane co-chaperone that interacts with BiP in dog microsomes (24) and stimulates its ATPase activity in vitro (23). Apart from the lumenal J-domain that mediates the interaction with BiP (23), MTJ1/ERdj1 contains a highly charged region adjacent to the transmembrane helix that binds to ribosomes and modulates the rate of translation of newly synthesized polypeptides (24). In addition, MTJ1/ERdj1 possesses a large carboxyl-terminal extension composed of two SANT domains and a spacer region for which the function was completely unknown. In the present study, we report the cloning of the human homologue HTJ1 (Fig. 1) and the critical role of its distal SANT domain (SANT2) in the formation of a stable complex with the serpin ACT. In another report, 2 we show that the HTJ1 SANT2 domain also interacts with ITIH4, a member of the inter-␣-trypsin inhibitor, reinforcing the concept that HTJ1 may be involved in multiple protein-protein interactions on both sides of the endoplasmic reticulum membrane. Its lumenal J-domain interacts with the molecular chaperone BiP (23), whereas its cytosolic domain may interact with ribosomes (24) and newly translated polypeptides.
The biochemical and kinetic studies presented in this article  show that SANT2 delays the formation of the ACT-chymotrypsin acyl complex (Fig. 9) and slows down the kinetics of association between ACT and chymotrypsin without affecting the catalytic properties of the enzyme (Fig. 10, Table II). A general model for the function of serpins based on their physical properties, mechanism, and structure has been established and recently described in fine detail (48,49). In addition to chymotrypsin, ACT inhibits the serine proteases cathepsin G, mast cell chymase, and proenkephalin processing enzymes (50 -53). ACT, like other serpins, is characterized by a mobile RSL that is a substrate for the protease, and a dominant ␤-sheet A that can open to insert the cleaved active site loop or bind exogenous free peptides of the same or similar sequence (50, 54 -56). Herein we present evidence that SANT2 binds to ACT within a region (residues 140 -343) that does not overlap with the RSL itself (Fig. 4) and that the resulting SANT2-ACT complex no longer exhibits inhibitory activity toward chymotrypsin (Figs. 8  and 9). The SANT2-ACT complex is very stable and forms with a dissociation constant equal to K app ϭ 0.5 nM for 4 nM ACT, a value that is about 4 order of magnitude smaller than the average serpin plasma concentration (around 45 mg/100 ml or 7 mM) in healthy individuals (57).
ACT is an abundant serum glycoprotein that is produced by liver cells through the secretory pathway. The ACT precursor contains a signal peptide (Fig. 3A) that is cleaved upon translocation across the ER membrane. Glycosylation occurs in the ER prior to transport to the Golgi apparatus and fusion of secretory vesicles with the plasma membrane. Interestingly, HTJ1 can rescue the thermosensitive phenotype of a Sec63 translocation-deficient mutant, 4 suggesting that HTJ1 functions in protein translocation across the endoplasmic reticulum membrane. Because a small positively charged region distal of the COOH-terminal of MTJ1 is associated with ribosomes (24), newly translated ACT may associate with HTJ1 through the SANT2 domain prior to its translocation across the ER. We attempt to detect ACT-HTJ1 complexes in membrane fractions prepared from non-transfected hepatocarcinoma HepG2 cells, or transfected with a vector that overexpresses MTJ1 (58). Unfortunately, ACT has a size similar to the heavy chain of immunoglobulins used in the immunoprecipitation experiments and could not be detected under these conditions. 5 It is thus unclear whether or not MTJ1/ERdj1 can stably interact with ACT in vivo. In mammalian cells, translocation is a cotranslational process and the complex between ERdj1 and ACT may be too short-lived to be detected.
A carboxyl-terminal fragment of MTJ1 was originally found in nuclear extracts (22), and localizes to the nucleus when expressed on its own in COS-7-transfected cells (58). The molecular mechanisms that generate the MTJ1 truncated nuclear form are unclear and remain under investigation in our laboratory. The Sec63-like region that separates the two SANT domains contains a putative nuclear localization signal (R 439 PRRRK, Fig. 1) that could target the soluble fragment to the nucleus. Interestingly, ACT also localizes in cell nuclei (59, 60) but does not possess any putative nuclear localization signal. ACT also has the unique property to bind DNA through exposed lysine residues with no effect on its protease inhibitory activity (61). ACT-DNA interaction is believed to play a role in tumor cell protection and in modulating the activity of chymotrypsin-like enzymes in chromatin (62). However, very little has been done in this area and the role of ACT DNA binding activity in the nucleus remains unknown. Interestingly, two recent reports (63,64) describe that a developmentally regulated serpin, the myeloid and erythroid nuclear termination stage-specific protein or MENT, also localizes to the nucleus and exerts a role in chromatin remodeling. It is very tempting to propose that the interaction of MTJ1 with ACT could mediate membrane release and nuclear localization of the complex and regulation of gene transcription through the DNA binding activity of ACT itself or through protein-protein interactions mediated by the HTJ1 SANT domain(s). Studies are underway to further characterize the role of HTJ1 in ACT secretion, function, and organelle targeting.
SANT domains are reminiscent of tryptophan-mediated DNA binding repeats and are found in many proteins, only few of which have been studied such as MIDA-1, a murine Jdomain protein involved in cell growth (65)(66)(67), the Zuotinrelated factors (68 -70), the proto-oncogene c-myb (71), the Mphase phosphoprotein 11 (72), Ada2 (33), and the histoneassociated CoREST and Mta-L1 (34). The arrangement of the two SANT domains in MTJ1, separated by a 116-residue spacer, is similar to that in CoREST (32,34), N-CoR, and SMRT (37). In the case of N-CoR and SMRT, the SANT domains not only interact with HDAC3 (histone deacetylase 3) but also activate its deacetylase activity (36,37). The yeast and human Ada2p are found in histone acetyltransferase complexes and their SANT domains are important for acetylation of histone NH 2 -terminal tails (33,35). Although highly homologous, SANT1 and SANT2 are not interchangeable and ACT has no affinity for SANT1. Although our screen was performed with the complete cytosolic region of HTJ1, no SANT1-interacting protein has been identified so far. Protein modeling of HTJ1 SANT1 and SANT2 domains indicate that although the overall folds are very similar, surface residues are much more polar in SANT1 than they are in SANT2. 2 These differences may be sufficient to control the specificity and thus the function of HTJ1 SANT domains. This hypothesis is currently being tested in our laboratory.
Aromatic residues are highly conserved in all SANT-containing proteins (30) and are important for their function. For example, substitution of tryptophans to alanine in Ada2 leads to a SAGA complex partially defective for nucleosome acetylation (33) and greatly reduces the sequence-specific DNA binding activity of c-myb (71). We clearly show here that tryptophan 520 is critical for the formation of the SANT2-ACT complex, whereas the other tryptophan residue (Trp-497) appears to be more closely engaged in the interactions with ITIH4. 2 In both cases, however, substitution of the other tryptophan by an alanine greatly decreases the affinity for the substrate. These observations indicate that the HTJ1 SANT2 domain can interact with several target proteins through overlapping but somehow distinct surface areas.