|Home | About | Journals | Submit | Contact Us | Français|
Site-specific proteolytic processing plays important roles in the regulation of cellular activities. The histone modification activity of the human trithorax group mixed-lineage leukemia (MLL) protein and the cell cycle regulatory activity of the cell proliferation factor herpes simplex virus host cell factor 1 (HCF-1) are stimulated by cleavage of precursors that generates stable heterodimeric complexes. MLL is processed by a protease called taspase 1, whereas the precise mechanisms of HCF-1 maturation are unclear, although they are known to depend on a series of sequence repeats called HCF-1PRO repeats. We demonstrate here that the Drosophila homologs of MLL and HCF-1, called Trithorax and dHCF, are both cleaved by Drosophila taspase 1. Although highly related, the human and Drosophila taspase 1 proteins display cognate species specificity. Thus, human taspase 1 preferentially cleaves MLL and Drosophila taspase 1 preferentially cleaves Trithorax, consistent with coevolution of taspase 1 and MLL/Trithorax proteins. HCF proteins display even greater species-specific divergence in processing: whereas dHCF is cleaved by the Drosophila taspase 1, human and mouse HCF-1 maturation is taspase 1 independent. Instead, human and Xenopus HCF-1PRO repeats are cleaved in vitro by a human proteolytic activity with novel properties. Thus, from insects to humans, HCF proteins have conserved proteolytic maturation but evolved different mechanisms.
Site-specific proteolysis has emerged as an important mechanism contributing to the regulation of basic cellular processes such as development (24), metabolism (3), cell cycle progression (7), and apoptosis (22). Site-specific proteolysis regulates both physiological and disease processes, often by activating latent functions.
In human cells, a number of nuclear proteins, such as the trithorax group mixed-lineage leukemia (MLL) protein (11, 21, 35) and herpes simplex virus (HSV) host cell factor 1 (HCF-1) (15, 31, 32), are synthesized as precursors that undergo proteolytic maturation to generate stable, noncovalently associated heterodimeric complexes. The importance of these proteolytic maturation processes is underscored by the finding that the Drosophila homologs of these proteins, Trithorax (Trx) and dHCF, also undergo proteolytic maturation (16, 18). Nevertheless, the cellular mechanism of proteolytic maturation is completely known only for human MLL.
MLL is cleaved by a novel endopeptidase called taspase 1 that utilizes an N-terminal threonine generated by autoproteolysis as the nucleophile for polypeptide cleavage (10). For HCF-1, autocatalytic processing of bacterially synthesized HCF-1 has been observed (28), but the mechanisms of HCF-1 maturation in human cells remain to be clarified. Curiously, in addition to sharing proteolytic maturation processes, MLL and HCF-1 bind each other (36), and both play important roles in the regulation of the cell division cycle (6, 23, 26, 30). These relationships encouraged us to compare their proteolytic maturation pathways along with those of the Drosophila Trx and dHCF homologs.
Of these four proteins, MLL is the largest, consisting of 3,969 amino acids. It was originally discovered because the gene encoding MLL is the site of chromosomal translocations in human childhood leukemias (1, 5, 8). MLL possesses histone H3 lysine 4 (H3K4) methyltransferase activity (20, 21) and is required for maintaining proper Hox and Cyclin gene expression (26, 37). MLL is cleaved by taspase 1 (10) at two sites to generate the associated amino-terminal (MLLN) and carboxy-terminal (MLLC) subunits (11, 21, 35). This proteolysis enhances the H3K4 methyltransferase activity of the MLLC subunit, which promotes Cyclin gene expression and cell cycle progression (26).
The enzyme responsible for Trx cleavage is not known, but it has been hypothesized that it is a homolog of taspase 1 because the region of cleavage (16) contains a putative taspase 1 recognition site (10, 35). The possible importance of Trx cleavage for its biological function has been indicated by the activity of a mutant Trx protein, called TrxE3, which contains a 271-amino-acid deletion (19) that spans the predicted processing site and abrogates Trx protein maturation (16). Trx is required to maintain proper expression of antennapedia and bithorax complex genes during fly development (2). Interestingly, trxE3 mutants display defective antennapedia but not bithorax complex gene expression (25), suggesting a selective role of Trx cleavage in its function.
HCF-1 is a 2,035-amino-acid chromatin-associated protein that was first discovered as a transcriptional coactivator for HSV immediate-early gene transcription, where it stimulates formation of the viral VP16-induced transcriptional regulatory complex (see reference 33 for a review). Proteolytic maturation of HCF-1 involves multiple cleavages at any one of six centrally located 26-amino-acid repeats called HCF-1proteolytic (HCF-1PRO) repeats, generating the associated amino-terminal (HCF-1N) and carboxy-terminal (HCF-1C) subunits (15, 31, 32). The HCF-1N and HCF-1C subunits play separate roles in two key phases of the cell cycle: the HCF-1N subunit promotes passage through the G1 phase, and the HCF-1C subunit is necessary for proper mitosis and cytokinesis during the M phase (14). Proteolytic processing is necessary to ensure proper HCF-1 function, as HCF-1C subunit functions are inhibited if the HCF-1 precursor cannot be processed (14).
Consistent with the importance of HCF-1 proteolytic maturation, the 1,500-amino-acid dHCF protein, although lacking HCF-1PRO repeats, also undergoes proteolytic maturation to generate associated dHCFN and dHCFC subunits (18). The dHCFN and dHCFC subunits display considerable structural similarity to the human HCF-1N and HCF-1C subunits (Fig. (Fig.1)1) as well as functional conservation. Thus, as for the HCF-1N subunit, the dHCFN subunit can associate with the HSV transactivator VP16 and stabilize the VP16-induced transcriptional regulatory complex (18), as well as associate with the Drosophila cell cycle regulators dE2F1 (the homolog of human E2F1) and dE2F2 (the homolog of human E2F4) (27).
Following our interest in HCF protein proteolytic maturation and in the conservation of HCF proteins in animals, we noted that dHCF, albeit not human HCF-1, contains a potential taspase 1 cleavage site within the region predicted by Mahajan et al. (18) for cleavage. This observation has led us to find that the dHCF and Trx proteins are both cleaved by the Drosophila taspase 1 homolog, whereas human HCF-1 proteolytic cleavage is independent of taspase 1. Thus, the Drosophila and human trithorax group proteins share proteolytic maturation pathways, while the Drosophila and human HCF proteins, although conserving proteolysis as a maturation process, mature via different proteolytic mechanisms.
A full-length Drosophila taspase 1 cDNA (LD05057) was obtained from the Drosophila Genomics Resource Center, Bloomington, IN, and cloned into the pET-28a(+) His-tagged bacterial expression vector (Novagen). The His-tagged full-length human taspase 1 cDNA bacterial expression vector and the vector for in vitro transcription/translation containing wild-type and CS1/2-mutated MLL amino acids 2400 to 2900 have been described previously (10). The full-length dHCF cDNA cloned in a T7 epitope-tagged Drosophila expression vector (pACXT-dHCF) was a kind gift of Angus Wilson (New York University). For in vitro transcription/translation of the precursor, a PCR fragment encoding dHCF amino acids 733 to 1212 was amplified from pACXT-dHCF and inserted in the pNCITE vector (30). For the Trx precursor, a PCR fragment encoding Trx amino acids 1973 to 2458 was amplified by reverse transcription-PCR from SL2 cell RNA, verified by sequence analysis, and inserted directly into the pGEM-T vector (Promega). Fragments consisting of HCF-1 amino acids 686 to 1166 derived from either wild-type or mutant pCGNHCFFL (31, 32) were PCR amplified using a forward primer including the phage T7 promoter and β-globin translational initiation codon and directly used for in vitro transcription/translation. The Xenopus HCF-1 (xHCF-1) precursor, corresponding to amino acids 1358 to 1653, was PCR amplified from pCMV-SPORT6-xHCF-11358-1653 (clone ID 7653830; BioCat) as for HCF-1 and directly used for in vitro transcription/translation. In vitro transcription/translation and [35S]methionine labeling were performed using the TNT T7 Quick transcription/translation system (Promega) as recommended by the manufacturer. Mutants of the Trx CS2-like and dHCF CS1-like sites were generated by QuickChange site-directed mutagenesis (Stratagene).
His-tagged human taspase 1 and Drosophila taspase 1 were expressed in Escherichia coli BL21(DE5) cells and purified by Ni affinity chromatography (QIAGEN) as recommended by the manufacturer. Proteins eluted from the resin were dialyzed against phosphate-buffered saline, 25% glycerol.
HeLa cells were grown at 37°C in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (FBS). Drosophila SL2 cells were grown at 25°C in Schneider's medium (GIBCO) with 10% heat-inactivated FBS. Nuclear and cytosolic extracts from HeLa and SL2 cells were prepared as previously described (4). Wild-type or taspase 1−/− mouse embryonic fibroblasts (MEFs) from day 12.5 embryos were grown as described previously (26). Extract was prepared by lysis in radioimmunoprecipitation assay buffer with Complete protease inhibitor cocktail (Roche) on ice for 30 min and clarified by centrifugation at 20,000 × g for 20 min.
Taspase 1 in vitro cleavage assays were performed as previously described (10). For comparative experiments, recombinant human taspase 1 and Drosophila taspase 1 activities were titrated on the respective cognate MLL and Trx substrates and used in the amount required to cleave 50% of the substrate (corresponding to 10 ng of human taspase 1 and 50 ng of Drosophila taspase 1). For the xHCF-1 in vitro cleavage assay, 1 μg of human taspase 1 and Drosophila taspase 1 were used. Cell extract in vitro cleavage assays were performed using 22 μl of the indicated cell extract in a 30-μl reaction mixture and incubating reaction mixtures at 30°C for 8 h or the indicated time period. HeLa cell extract heat treatment was at 65°C for 20 min, and protease inhibitors were utilized at the final concentration of 0.5 mg/ml Pefabloc (Roche) and 1× Complete protease inhibitor cocktail (Roche). The reaction mixtures were resolved by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), the resulting radioactive signals were visualized with a Typhoon Trio+ imaging system and quantified with ImageQuant (Amersham Biosciences).
A 183-bp double-stranded RNA (dsRNA) (RNAi 1) and a 495-bp dsRNA (RNAi 2) against the Drosophila taspase 1 gene sequence as well as a nonspecific firefly luciferase dsRNA were used for RNA interference (RNAi). dsRNAs were synthesized with a MEGAscript T7 kit (Ambion), using as templates PCR products amplified with primer pairs (RNA: 2), CAGTCGTTCTGTGTCGGCTA and CTAGGAGCTGCAGGAAGGTG (RNA: 1) and TGGGTTGCTCCATGGTG and TCCAGCGTACAGATCCTG each containing an additional 5′ T7 promoter sequence. For dsRNA treatment, 1 × 106 SL2 cells were seeded in six-well plates in 1 ml of serum-free Schneider's medium and incubated at 25°C for 2 h in the presence of 15 μg of dsRNA before addition of 2 ml of Schneider's medium with 10% heat-inactivated FBS. dsRNA-treated cells were collected after 96 h. For transient transfection, 2 × 106 untreated or 48-h-dsRNA-treated SL2 cells were transfected with 5 μg of DNA using Cellfectin (Invitrogen) as described by the manufacturer. Cells were collected at 48 h posttransfection. For collection, SL2 cells were washed with phosphate-buffered saline, lysed in Laemmli SDS sample buffer, and subsequently analyzed by immunoblot analysis.
Protein lysates were resolved by SDS-PAGE, transferred to nitrocellulose membrane, and probed with T7 tag monoclonal antibody (Novagen), polyclonal HCF-1C antiserum (H12) (31), or polyclonal dHCFN antiserum, followed by appropriate IRDye800-conjugated anti-mouse or IRDye680-conjugated anti-rabbit immunoglobulin G (Rockland). The immunoreactive bands were detected by fluorescence with an Odyssey infrared imager (LI-COR). The polyclonal dHCFN antiserum was raised in rabbits against a 15-amino-acid peptide (EGSDFVDPAFSSGER) corresponding to the N terminus of dHCF and affinity purified with the peptide immunogen.
Figure Figure11 shows schematic representations of the human and Drosophila MLL/Trx and HCF proteins highlighting conserved sequence and structural elements. Figure Figure1A1A also shows the positions of the two known taspase 1 cleavage sites in MLL, called CS1 (QVD/GADD) and CS2 (QLD/GVDD) (10, 11, 35). By sequence similarity, Trx contains a CS2-like site (QMD/GVDD) (10, 11, 35) and dHCF contains a CS1-like site (QLD/GAGD), both of which lie in positions analogous to the cleavage sites in the respective human homologs (i.e., MLL CS1 and CS2 sites and HCF-1PRO sites). We therefore tested whether the CS2-like Trx and CS1-like dHCF taspase 1 recognition sites are substrates for taspase 1.
In initial experiments, we tested the ability of human taspase 1 to cleave the Trx and dHCF proteins and found that the human protease showed a strong preference for the human MLL substrate (see below). Therefore, we tested the ability of the Drosophila taspase 1 homolog to cleave Trx and dHCF substrates as shown in Fig. Fig.2.2. To facilitate this analysis, we generated for each protein shorter precursor cleavage substrates that contain the regions with the putative taspase 1 cleavage sites. Both wild-type and cleavage site mutant substrates were used (Fig. (Fig.2A).2A). These precursors were synthesized in vitro in the presence of [35S]methionine and incubated with recombinant Drosophila taspase 1 purified from E. coli, and the products were separated by SDS-PAGE. Figure Figure2B2B shows that Drosophila taspase 1 can cleave the wild-type Trx precursor (lanes 1 and 2) but not the precursor carrying the mutation in the CS2-like site (lanes 3 and 4). These results suggest that, like its human homolog MLL, Trx is cleaved by taspase 1 at the CS2-like cleavage site. Similarly, as shown in Fig. Fig.2C,2C, Drosophila taspase 1 can effectively cleave the wild-type dHCF precursor (lanes 1 and 2) but not the CS1-like mutant precursor (lanes 3 and 4). These experiments define the first heterologous protease for HCF protein maturation, the Trx protease Drosophila taspase 1.
To characterize this dHCF protease further, we assessed whether dHCF is cleaved at the taspase 1 cleavage site in vivo. We transfected Drosophila SL2 cells with an epitope-tagged expression vector encoding full-length wild-type and CS1-like mutant recombinant dHCF (rdHCF) and assayed cleavage by immunoblotting with an antitag antibody. As shown in Fig. Fig.2D,2D, synthesis of wild-type rdHCF resulted in an approximately 105-kDa N-terminal fragment (rdHCFN) (lane 2) (18), whereas synthesis of the CS1-like mutant rdHCF led to an approximately 180-kDa product, a size equivalent to that of the full-length protein (rdHCFFL) (lane 3) (18). These results indicate that dHCF is cleaved at a single taspase 1 recognition site.
We also analyzed proteolytic maturation of endogenous dHCF by generating an antibody directed to the N terminus of the protein (see Materials and Methods). Immunoblot analysis of the same SL2 extracts shown in Fig. Fig.2D2D revealed that endogenous dHCFN comigrates with the processed rdHCFN (compare lanes 1 and 2 in Fig. 2D and E). With this antibody, we could not detect any endogenous full-length dHCF protein, but we could detect the ectopic CS1-like mutant rdHCFFL (compare lane 3 with lane 1). The lack of endogenous full-length dHCF indicates, as previously shown using a dHCFC antibody (9), that dHCF is efficiently processed in Drosophila cells.
To determine whether Drosophila taspase 1 is the sole protease responsible for dHCF processing in vivo, we depleted Drosophila taspase 1 from Drosophila cells by RNAi. As shown in Fig. Fig.2F,2F, independent treatment of SL2 cells with two different taspase 1 dsRNAs (RNAi 1 and RNAi 2) led in each case to the detection of dHCFFL (lanes 3 and 4) compared to untreated (lane 1) or mock RNAi-treated (luciferase, lane 2) cells. The levels of dHCFFL were limited, however, possibly owing to high levels of stable dHCFN remaining from before the RNAi treatment. In contrast, we observed prominent inhibition of rdHCF cleavage when the RNAi-induced Drosophila taspase 1 depletion was associated with concomitant synthesis of dHCF by simultaneous transfection of the epitope-tagged dHCF expression vector. Thus, as shown in Fig. Fig.2G,2G, in taspase 1 RNAi-treated cells, rdHCF processing is significantly reduced (lanes 3 and 4) compared to that in untreated (lane 1) or mock RNAi-treated (lane 2) cells. In summary, our in vitro and in vivo results indicate that Drosophila taspase 1 is responsible for dHCF processing at a single site and indeed may be the sole protease responsible for dHCF maturation. Thus, in flies, where dHCF lacks HCF-1PRO repeats, HCF protein maturation is apparently not autocatalytic as observed for an HCF-1PRO repeat region of human HCF-1 synthesized in E. coli (28).
We were surprised to find that dHCF is a substrate of Drosophila taspase 1, because Izeta et al. (12) have shown that dHCF is not processed in hamster cells and yet such cells would be expected to possess taspase 1. To explore this apparent discrepancy, we compared directly the abilities of human and Drosophila cell extracts to cleave the dHCF precursor. Indeed, as the results of Izeta et al. (12) would suggest, the dHCF precursor, albeit cleaved by a Drosophila SL2 extract, was not effectively cleaved by a human HeLa cell extract, as shown in Fig. Fig.3A3A (compare lanes 2 and 3). This dissimilarity is likely the result of intrinsic differences in the human taspase 1 and Drosophila taspase 1 enzymes, because the same species-specific activity was also observed with purified recombinant human taspase 1 and Drosophila taspase 1 (compare lanes 5 and 6) (see Materials and Methods for enzymatic activity normalization). These results explain why Izeta et al. (12) did not observe dHCF processing by using mammalian cells and emphasize the importance of using a homologous system to assay HCF protein processing (18).
The dHCF Drosophila taspase 1 and human taspase 1 cleavage results suggest that taspase 1 activity has changed during evolution. To study this evolution further, we compared the human taspase 1 and Drosophila taspase 1 enzymes on homologous substrates: precursors from the human MLL and Drosophila Trx proteins (Fig. (Fig.3B).3B). As shown in Fig. Fig.3C,3C, a human HeLa cell and a Drosophila SL2 cell extract display enhanced activity on the cognate MLL and Trx precursor, respectively (compare lanes 2 and 3 with lanes 5 and 6) (in this assay only MLL-CS2 cleavage is indicated). Furthermore, purified recombinant human taspase 1 and Drosophila taspase 1 display the same cognate substrate preference (Fig. (Fig.3D,3D, compare lanes 2 and 3 with lanes 5 and 6). These results suggest that the taspase 1 enzymes have coevolved with their substrates MLL and Trx.
Zhou et al. (38) have defined a human taspase 1 recognition heptapeptide sequence. The MLL CS2 and Trx CS2-like sites differ by a single amino acid (from QLDGVDD in MLL CS2 to QMDGVDD in Trx). To examine the determinants responsible for the cognate enzyme substrate preference of Drosophila taspase 1 on Trx, we converted the Trx CS2-like taspase 1 site to the MLL CS2 sequence via a single amino acid change (Fig. (Fig.3B).3B). As shown in Fig. Fig.3E,3E, the humanized Trx precursor [Trx(MLLCS2)] is still a better substrate for Drosophila taspase 1 than human taspase 1, indicating that the species-specific taspase 1 selectivity observed here results from sequence differences that lie outside the so-far-identified heptapeptide recognition sequence.
HCF-1 is cleaved at any one of the six HCF-1PRO repeats (15, 31), and when the region containing all six sites is deleted (14, 32) or the six sites are all individually inactivated by amino acid substitution (29), the protein is no longer cleaved. To study HCF-1 proteolytic cleavage, we first asked whether the HCF-1PRO repeat is a taspase 1 substrate. Therefore, we prepared an in vitro HCF-1 protease substrate containing three of the six HCF-1PRO repeats (HCF-1PRO repeats 1, 2, and 3, called HCF-1rep123) (Fig. (Fig.4A)4A) and incubated it with recombinant human taspase 1. Figure Figure4B4B shows that human taspase 1 had no effect on this HCF-1 precursor (compare lanes 1 and 2) at a concentration that effectively cleaved the MLL precursor (compare lanes 5 and 6), suggesting that taspase 1 is not an HCF-1PRO repeat protease.
To further exclude the requirement of taspase 1 in HCF-1 maturation, we asked whether HCF-1 processing is affected by the absence of taspase 1. Takeda et al. (26) have shown using MEFs that MLL processing is disrupted in taspase 1−/− knockout mice. In contrast, as shown in Fig. Fig.4C,4C, HCF-1 is processed normally in both wild-type and taspase 1−/− MEFs (compare lanes 2 and 3), with the pattern of cleavage products being identical to the one observed in a HeLa cell extract (lane 1). Together, these results indicate that taspase 1 is not the protease responsible for HCF-1PRO repeat processing. Indeed, taspase 1 is unlikely to be involved in HCF-1 processing generally, as there is no identifiable taspase 1 cleavage site in the entire HCF-1 amino acid sequence.
To identify an activity responsible for HCF-1PRO repeat processing, we used a wild-type HCF-1rep123 substrate or substrates with HCF-1PRO repeats 1 and 2 mutated either individually (HCF-1repX23 and HCF-1rep1X3) or together (HCF-1repXX3) as illustrated in Fig. Fig.5A.5A. As shown in Fig. Fig.5B,5B, in the absence of cell extract, none of the precursors was effectively cleaved (lanes 1, 3, 5, and 7). In the presence of HeLa cell extract, however, each precursor was cleaved at the wild-type but not mutated HCF-1PRO repeats (compare lanes 2, 4, 6, and 8), displaying a cleavage pattern consistent with HCF-1PRO repeat specific cleavage. These results suggest that HeLa cell extracts possess an activity that can specifically cleave an HCF-1PRO repeat precursor.
To establish the nature of this HCF-1PRO repeat enzymatic activity, we performed a time course experiment using the HCF-1rep123 precursor. As shown in Fig. Fig.5C,5C, under the same assay conditions used for Fig. Fig.5B,5B, the HCF-1PRO repeat cleavage product increased with incubation time for approximately 8 h. Accumulation of the product resulting from cleavage at the HCF-1PRO repeat 1 was quantified as shown in Fig. Fig.5D.5D. These results suggest that in cell extracts the HCF-1PRO repeat proteolytic cleavage is relatively slow but stable for 8 h.
To characterize the HCF-1PRO repeat protease further, we assayed the sensitivity of the HeLa cell extract activity to heat and protease inhibitor treatment as shown in Fig. Fig.5E.5E. The HeLa cell activity was sensitive to heat treatment (compare lanes 1 to 3) but resistant to Pefabloc, a serine protease inhibitor (compare lanes 3 and 4) (an assay with thrombin and a substrate with a thrombin cleavage site showed that the serine protease inhibitor was active [data not shown]), and Complete, a serine and cysteine protease inhibitor cocktail (compare lanes 3 and 5). Thus, the HCF-1PRO repeat protease displays both sensitivity (heat) and resistance (serine and cysteine protease inhibitors) to various treatments. We note that the resistance of the HCF-1PRO repeat protease to a serine protease inhibitor differentiates it from the autocatalytic activity described by Vogel and Kristie, which was serine protease inhibitor sensitive (28).
The identification of taspase 1 as the protease responsible for dHCF maturation and the observation that human HCF-1 is cleaved by a different activity at the HCF-1PRO repeats indicate that, from fly to human, HCF proteins have conserved proteolytic maturation but evolved different mechanisms. These differences between flies and humans appear to be generally specific to insects or vertebrates, as illustrated in Fig. Fig.6.6. Thus, the Apis mellifera (honeybee) HCF protein possesses a consensus taspase 1 cleavage site at the same relative location as in the dHCF protein (Fig. (Fig.6A),6A), suggesting that taspase 1 cleavage of HCF proteins may be generally conserved in insects. In parallel, a comparison of human, frog, and fish HCF-1 proteins shows that the position (Fig. (Fig.6B)6B) and sequence (Fig. (Fig.6C)6C) of the HCF-1PRO repeats have been very highly conserved in these three distantly related vertebrate species. Thus, between insects and vertebrates, there appears to have been an evolutionary switch in HCF protein processing mechanism that has been highly conserved within each group.
Interestingly, in some fish (e.g., F. rubripes) and frog (e.g., X. tropicalis) species, in addition to the perfectly conserved HCF-1PRO repeats there is a partially conserved taspase 1 cleavage site (six-of-seven match to a heptapeptide consensus taspase 1 cleavage site) (38) located just C terminal of the last HCF-1PRO repeat (Fig. (Fig.6B).6B). This partially conserved taspase 1 cleavage site could represent an active site, indicating that in some species HCF-1 could be processed by both taspase 1 and an HCF-1PRO repeat protease. Alternatively, it could be inactive and instead represent the vestige of an active taspase 1 cleavage site in an ancestral HCF-1 protein. To distinguish between these two possibilities, we asked whether an xHCF-1 cleavage precursor containing the partially conserved taspase 1 cleavage site and the adjacent two HCF-1PRO repeats (xHCF-1rep89 [Fig. [Fig.6D])6D]) could be a substrate for either taspase 1 and/or the HCF-1PRO repeat protease. As shown in Fig. Fig.6D,6D, the xHCF-1rep89 substrate is not cleaved by either Drosophila taspase 1 or human taspase 1 at concentrations that can effectively cleave the dHCF precursor (compare lanes 1 to 3 and 4 to 6). Instead, as for human HCF-1 (lanes 8 and 9), the xHCF-1rep89 precursor is cleaved by the HeLa cell extract at positions corresponding to the HCF-1PRO repeats (compare lanes 4 and 7). These data suggest that, like mammalian HCF-1 proteins, xHCF-1 is not cleaved by taspase 1 but is instead processed at the HCF-1PRO repeats by the same activity responsible for human HCF-1 maturation. The ability of the human cell extract to cleave the Xenopus HCF-1PRO repeats provides functional evidence of the high degree of conservation of the HCF-1PRO repeat proteolytic process in vertebrates.
In this study, we have compared the proteolytic maturation processes of two pairs of evolutionarily conserved proteins: the trithorax group proteins human MLL and Drosophila Trx and the HCF proteins human HCF-1 and dHCF. We have shown that, like MLL, the Drosophila Trx is a substrate for the taspase 1 protease. Interestingly, the enzyme-substrate specificity, i.e., human taspase 1 with MLL and Drosophila taspase 1 with Trx, has coevolved. In contrast, with respect to HCF proteins, very different proteolytic pathways have evolved. Thus, the dHCF protein is cleaved by Drosophila taspase 1 but its human (and Xenopus) HCF-1 counterparts are cleaved by what is likely an unusual proteolytic activity.
Hsieh et al. (10) noted that sequences encoding the taspase 1 enzymes have been conserved in vertebrates and insects but not in the worm Caenorhabditis elegans and that, correspondingly, the MLL/Trx developmental regulators in vertebrates and insects but not in worms have taspase 1 recognition sites. Consistent with this observation, Trx is indeed a taspase 1 substrate at its CS2-like taspase 1 cleavage site. Examination of the activities of the human and Drosophila taspase 1 enzymes on the MLL and Trx proteins also revealed a more refined coevolution of enzyme and substrate, as human taspase 1 is more active on its cognate MLL substrate and Drosophila taspase 1 is more active on its cognate Trx substrate. This observation suggests that the presence of the taspase 1 enzyme and MLL/Trx protein maturation have been conserved because the cleavage is critically important for proper development (16, 19, 26). Nevertheless, there is considerable flexibility in the interaction between protease and substrate, and for such enzyme-substrate coevolutionary flexibility to be possible, it is likely that taspase 1 does not possess a large number of essential targets.
We note with interest that vertebrates and insects display common longitudinal body segmentation, which is lacking in C. elegans, and that proper MLL and Trx maturation is functionally linked to proper homeodomain gene expression, which is intimately involved in segmentation determination (10, 16, 19, 26). Consistent with the suggestion that MLL/Trx cleavage is required to control the expression of genes involved in body segmentation, in the sea urchin Strongylocentrotus purpuratus, which, like C. elegans, is not longitudinally segmented, the Trx homolog lacks an evident taspase 1 recognition site and there is no evident taspase 1-encoding gene (data not shown).
Although HCF proteins are not known to be involved in regulating segmentation, invertebrate HCF proteins display species-specific patterns of proteolytic maturation that follow the Trx proteins. Thus, in Drosophila, dHCF and Trx are both cleaved by Drosophila taspase 1, and in the honeybee the HCF and Trx homologs possess taspase 1 recognition sites. In contrast, as with the Trx homologs, the sea urchin and worm HCF proteins do not possess evident taspase 1 recognition sites (data not shown); indeed, worm HCF does not undergo proteolytic maturation (34). We suggest that in invertebrates, the HCF and Trx proteins coevolved taspase 1-induced maturation. In contrast, in vertebrates, HCF-1 proteins took on a very different proteolytic maturation mechanism. Nevertheless, the parallels between MLL/Trx and HCF protein maturation suggest that in addition to regulation of the cell cycle, HCF-1 may also play a role in regulation of genes involved in segment determination.
We have described an enzymatic activity in HeLa cell extracts that specifically cleaves the HCF-1PRO repeat. We do not know at this time the precise relationship between the HeLa cell activity described here and the HCF-1 autocatalytic activity described previously (28), but we note some differences between the two activities. The autocatalytic activity, which depends on sequences within the HCF-1C region, results in cleavage at not only HCF-1PRO repeats but also an additional site, called 102, which lacks evident HCF-1PRO repeat sequence similarity and is located just C terminal of the sixth HCF-1PRO repeat (15, 28). This autocatalytic site is, however, not used in vivo when the HCF-1PRO repeats are inactivated by mutation or deletion (14, 29, 32), and we have not observed 102 site cleavage using a precursor substrate spanning this site in the HeLa cell extract assay (F. Capotosti and W. Herr, unpublished results). Thus, in both HeLa cells and HeLa cell extracts, HCF-1 proteolytic maturation appears to be HCF-1PRO repeat specific. We also note that, unlike the autocatalytic 102 site activity, which is sensitive to the serine protease inhibitor Pefabloc (28), the HeLa cell HCF-1PRO repeat activity is resistant to this protease inhibitor (Fig. (Fig.5E).5E). Clarifying the relationship between the HeLa cell activity described here and the HCF-1 autocatalytic activity will require further investigation.
Whatever the relationship, however, the properties of the HCF-1PRO repeat protease are likely to be unusual. As shown in Fig. Fig.6C,6C, there are 16 amino acid residues over 19 positions that are identical among the HCF-1PRO repeat consensus sequence of the fish, frog, and human HCF-1 proteins, an impressive level of sequence conservation for a proteolytic recognition sequence. Furthermore, as described previously (32), alanine substitutions at 12 positions over 18 of the human HCF-1PRO repeat 2 affect HCF-1PRO repeat cleavage in HeLa cells (Fig. (Fig.6C),6C), a surprisingly large sequence requirement for proteolytic cleavage. Given this extensive sequence conservation and sequence requirements, we suggest that the functional HCF-1PRO repeat element reflects the assemblage of different recognition sequences for (i) the protease itself and (ii) accessory factors that could regulate proteolysis. Such accessory factors could directly recruit or stimulate the protease or could reflect enzymes that modify the HCF-1PRO repeat (e.g., by phosphorylation or glycosylation) to regulate cleavage. Whatever the reason, the large size of the HCF-1PRO repeat results in considerable specificity, as we have been unable to find a match to the HCF-1PRO repeat in any other protein sequence found in public protein databases, suggesting that in toto the mechanisms of HCF-1PRO repeat cleavage are unique to HCF-1 maturation.
The evolutionary comparisons shown in Fig. Fig.66 indicate that although the mechanisms of HCF protein maturation have switched between insects and vertebrates, the end results of proteolytic processing are similar. To illustrate this point, Fig. 6A and B show charge profiles for the two insect (fly and honeybee) and three vertebrate (fish, frog, and human) HCF proteins. These profiles show that the insect and vertebrate HCF proteins all contain basic and acidic regions at corresponding positions. This conservation of amino acid composition is consistent with the important cellular functions these regions have been shown to possess in human HCF-1: G1 phase progression in the case of the basic region (30) and M phase progression (13), transcriptional activation (17), and chromatin association (13) in the case of the acidic region. We note with interest that whether the HCF protein is cleaved by taspase 1, as appears to be the case with the insect HCF proteins, or at the HCF-1PRO repeats, as appears to be the case in the vertebrate proteins, the cleavage site(s) is always positioned between the basic and acidic regions. Thus, although the mechanism for HCF protein proteolysis has changed during evolution, the resulting HCFN and HCFC subunits are very similar in structure.
The unexpected switch in proteolytic processing mechanism between insect and vertebrate HCF proteins leads to the questions of how and why. Concerning how the HCF-1PRO repeats may have arisen, we note with interest that the six HCF-1PRO repeats in fish and human HCF-1 are all encoded by a single large exon of 1,477 bp in human. (The nine repeats in X. tropicalis are encoded by two exons [five in the first and four in the second], which may have resulted from a duplication of the six-repeat-containing fish/human exon [Fig. [Fig.6C].)6C].) We imagine that an HCF-1 progenitor acquired the six HCF-1PRO repeats as a single genetic element by recombination, perhaps transposition, prior to vertebrate evolutionary divergence.
Concerning why a switch may have occurred, one possible explanation is that the region containing the HCF-1PRO repeats has been evolutionarily selected as an additional platform for protein-protein interactions, as Vogel and Kristie (29) have shown that the transcriptional coactivator/corepressor FHL2 interacts with nonprocessed HCF-1, stimulating transcription of an HCF-1 target gene. This interaction between HCF-1PRO repeats and cofactors could allow the modulation of HCF-1 processing and activity. Another possible explanation is that, with an HCF-1PRO repeat protease, HCF-1 protein maturation has become independent from taspase 1 and thus from MLL/Trx maturation. This could result in a more flexible regulation of these two important cell cycle regulators. In any case, however, the importance of HCF-1PRO repeat processing during vertebrate evolution is underscored by the remarkable similarity among the repeats themselves in one species (e.g., human) and between vertebrates as divergent as fish and human (Fig. (Fig.6C6C).
In conclusion, in human and fly, the MLL/Trx and HCF proteins have conserved the process of proteolytic maturation, but the MLL/Trx process coevolved with the cognate taspase 1 proteases, whereas HCF-1 and dHCF have apparently evolved very different proteolytic pathways. Whatever the reason for the evolutionary change to HCF-1PRO repeat-dependent processing in vertebrates, it is evident that the HCF-1PRO repeats are unusual cleavage sites, and it is likely that the protease responsible for their cleavage has unusual properties.
We thank Angus Wilson for the dHCF expression vector; Deborah Aufiero for help with characterizing the anti-dHCF antisera; Nouria Hernandez, Christina Hertel, Sara Rodriguez Jato, and Shweta Tyagi for critical readings of the manuscript; and Henrik Kaessmann for advice.
F.C. is the recipient of a Roche Research Foundation Fellowship. These studies were supported by the Swiss National Science Foundation and the University of Lausanne.
Published ahead of print on 13 August 2007.