A 69-Base Pair Fragment Derived from Human Transcobalamin II Promoter Is Sufficient for High Bidirectional Activity in the Absence of a TATA Box and an Initiator Element in Transfected Cells

A 69-base pair (bp) (−581/−513) fragment derived from human transcobalamin II distal promoter constructed upstream of a chloramphenicol acetyltransferase reporter gene demonstrated high bidirectional promoter activity in transfected epithelial Caco-2 cells. DNase I footprinting, gel mobility shift, supershift, and mutagenesis studies with the 69-bp fragment demonstrated that a GC box (−568/−559) and an E box (−523/−528), which interacted with Sp1/Sp3 and USF1/USF2 (where USF is upstream stimulatory factor), respectively, were required for the full transcriptional activity of this fragment. Whereas mutations in the GC box reduced the promoter activity by 50%, mutations in the E box alone or in both the E box and GC box resulted in 90% loss of transcriptional activity. The essential role of the E box in the bidirectional promoter activity was further demonstrated by transient transfection in Caco-2, K-562, and HeLa cells using a 29-bp (−541/−513) fragment that contained only the E box. Based on these results we suggest that 1) the E box is essential for both the GC box-dependent and -independent promoter activity of the 69-bp fragment, 2) cooperative interactions between Sp1/Sp3 and USFs are required for the full activation of the 69-bp promoter activity, and 3) the single E box is able to mediate bidirectional transcription in transfected cells in the absence of an obvious TATA box or a known initiator element.

referred to as class A, the E box with the sequence CACGTG belongs to class B. The functions of the E box are very divergent and to a large extent dependent upon the transcription factors that bind to it. These include neurogenesis, myogenesis, sex determination, T-cell/B-cell, and pancreatic specific gene expression, as well as cell proliferation and differentiation (reviewed in Ref. 2). The selective binding mechanism by which multiple factors bind to the same target site (class A or B) is not clear. It may involve competition among the factors influenced by sequences flanking the consensus CANNTG or interactions with other proteins binding to adjacent sites (3).
Previously (23) we have shown that the promoter activity of human transcobalamin II (TC II), a plasma transporter of cobalamin (24), is relatively weak and controlled positively by a distal GC box and negatively by a proximal GC/GT overlapping box. In the present study, we have identified a 69-bp sequence (Ϫ581/Ϫ513) from the distal region of the TC II promoter (25) that did not contain an obvious TATA box or a known Inr element but possessed a high promoter activity in transfected cells in an orientation-independent manner. The bidirectional promoter activity was due to interactions between Sp1/Sp3 and USF1/USF2 that bound to GC box and E box, respectively. Furthermore, a 29-bp sequence (Ϫ541/Ϫ513) containing only the E box was sufficient by itself to mediate transcription in the absence of other cis-elements. This finding suggests that TATA Ϫ /Inr Ϫ promoters may use an E box as a core element to direct basal transcription.

Construction of Promoter-CAT Reporter Plasmids-Promoterless
plasmid, pCAT-Basic (pCAT-B) (Promega, Madison, WI), was used for the preparation of the CAT fusion reporter constructs. Various truncated TC II-promoter fragments were generated by polymerase chain reaction (PCR) and inserted into the pCAT-B vector at a PstI site upstream from the CAT gene. A total of seven promoter fragments were amplified, and the sequences of each pair of primers used for PCR are shown in Table I. The DNA sequence of each of the promoter fragments was confirmed by sequencing prior to transfection. For promoter fragments CI (25-bp) and CII (29-bp), double-stranded (DS) oligonucleotides corresponding to the respective sequences (Table II) were synthesized and ligated at a PstI site in front of the CAT gene in the pCAT-B vector. In addition, fragments F2 and CII were also cloned into the pCAT-Promoter (pCAT-P)(Promega) vector at a site about 2.8 kilobase pairs upstream of the start site of the CAT gene.
Site-directed Mutagenesis-The substitution mutagenesis of the GC or E box in the promoter fragment was generated using the Quick-Change Site-directed mutagenesis method (Stratagene, La Jolla, CA). For each mutagenesis a pair of overlapping oligonucleotides with desired mutations was commercially synthesized, and the sequences are listed in Table II. The mutations generated were confirmed by DNA sequencing.
Cell Culture and Transient Transfection-The human colon epithelial cell line Caco-2 cells were obtained from American Type Culture Collection (Rockville, MD) and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. The cells were plated at a density of 1.5 ϫ 10 6 cells/100-mm dish the day before transfection and transfected by the calcium phosphate precipitation method (26). The transfection was performed using 15 g of promoter-CAT fusion plasmid and 5 g of an internal control plasmid pSV-␤-galactosidase (Promega) for monitoring the transfection efficiency. In addition, promoterless vector pCAT-B and promoter-containing plasmid pCAT-P were also transfected in each experiment as a negative and positive control, respectively. Each transfection was carried out at least three times.
␤-Galactosidase and CAT Assays-After 46 h of transfection, cells were harvested, and extracts were prepared for ␤-galactosidase and CAT assays. ␤-Galactosidase activity was measured according to Herbomel et al. (27), and CAT activity was determined by the method of Gorman et al. (28) using [ 14 C]chloramphenicol as a substrate. The acetylated products were separated from the unacetylated products by either thin layer chromatography or liquid scintillation counting as described by Seed and Sheen (29). CAT activities were normalized to ␤-galactosidase activities for the variations of transfection efficiency.
Preparation of Nuclear Extracts-Nuclear extracts were prepared from Caco-2, HeLa, and K-562 cells essentially as described by Dignam et al. (30). All buffers contained protease inhibitors including phenylmethanesulfonyl fluoride (1 mM), leupeptin (2 g/ml), antipain (10 M), and benzamidine (1 mM). Protein concentration of the nuclear extract was determined by the Bio-Rad protein assay method using bovine serum albumin as a standard (31).
Electrophoretic Mobility Shift Assay (EMSA)-The promoter fragment C (Ϫ581/Ϫ513) was labeled with [ 32 P]dCTP using the Klenow fragment of DNA polymerase. DS oligonucleotide CI or CII (Table II) was labeled at the 5Ј termini with [␥-32 P]ATP using T4 polynucleotide kinase. Labeled probe (ϳ2 ϫ 10 4 cpm) was incubated for 15 min at 22°C with 2.2 g of nuclear extract in 10 l of reaction buffer (10 mM Tris-HCl, pH 7.5, 4% glycerol, 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, and 50 g/ml poly(dI⅐dC)⅐poly(dI⅐dC)). For competition experiments the nuclear extract was incubated with the indicated concentrations of DS oligonucleotide competitor (Table II) at 22°C for 5 min prior to incubation with the probe. For immunosupershift assay, the nuclear extract was preincubated with 1.5 g of affinity purified rabbit polyclonal antibody against USF1, USF2, or c-Myc (Santa Cruz Biotechnology, Inc, Santa Cruz, CA) at 22°C for 30 min prior to incubation with the probe. The reaction mixture was then subjected to 4% polyacrylamide gel electrophoresis in 0.5ϫ TBE buffer (44.5 mM Tris-HCl, 44.5 mM boric acid, and 1 mM EDTA) at 100 V. The protein-DNA complexes were visualized by autoradiography.
DNase I Footprinting Analysis-The coding strand of the promoter fragment C was 3Ј-end-labeled by [ 32 P]dCTP using the Klenow fragment of DNA polymerase and gel-purified. The end-labeled probe (ϳ1 ϫ 10 4 dpm) was incubated for 10 min on ice with 20 g of nuclear extract in 50 l of buffer (25 mM Tris-HCl, pH 8.0, 50 mM KCl, 6.25 mM MgCl 2 , 0.5 mM EDTA, and 10% glycerol). After the incubation, 50 l of 5 mM CaCl 2 , 10 mM MgCl 2 solution was added to the reaction followed by digestion with 0.4 -2 units of DNase I (Promega) at 22°C for 2 min. The digestion was then terminated by adding 90 l of stop solution (200 mM NaCl, 30 mM EDTA, 1% SDS, and 100 g/ml yeast tRNA) and subjected to phenol/chloroform extraction. After precipitation with ethanol, the DNA pellet was resuspended in 4 l of the loading buffer, heated at 95°C for 2 min, and subjected to electrophoresis in a 6% polyacrylamide sequencing gel.

RESULTS
A 69-bp Sequence Derived from the 5Ј-Region of the TC II Gene Activates Transcription in an Orientation-independent but Position-dependent Manner-Our previous 5Ј-deletion studies (23) have demonstrated that the transcriptional activity of the TC II-promoter fragments, including the longest (Ϫ1014 to ϩ34), is very weak in Caco-2 cells. To identify regions that may potentially have higher promoter activity, four pro- moter fragments (F1-F4) generated by PCR were cloned into a promoterless CAT reporter vector, pCAT-B. Upon transient transfection in Caco-2 cells (Fig. 1A), only the fragment F2-(Ϫ746/Ϫ513) possessed promoter activity above the background activity produced by the promoterless pCAT-B vector. The promoter activity of this fragment (20-fold) was nearly 2.5-fold higher than that obtained using the pCAT-P vector (8.5-fold) that contained the SV40 promoter upstream from the CAT gene. In addition, the promoter activity of the F2 fragment was orientation-independent ( Fig. 1B) as the fragment F2 in a reverse orientation (Ϫ513/Ϫ746) revealed a similar level (19fold) of transcriptional activity.
In order to identify the elements within the 234-bp fragment that were responsible for its high bidirectional transcriptional activity, this fragment was further dissected into three regions (A, B, and C) (Fig. 1B). The 5Ј-terminal region A (58 bp, Ϫ746/ Ϫ689), which contained an inverted CCAAT element and a potential AP2-binding site, led to a 3-fold activation of the CAT activity in both orientations, accounting for about 16% of the activity of the fragment F2. The region B (119 bp, Ϫ688/Ϫ570) which contained several potential cis-elements, such as two AP2 sites, two Myb sites, and four inverted GA or GT boxes (GGGA/TGGG), did not reveal any promoter activity in either orientation. However, the 3Ј-terminal region C (69 bp, Ϫ581/ Ϫ513) had a high transcriptional activity (ϳ15-fold) in both orientations, corresponding to about 75% of promoter activity of the full-length F2 fragment.
Although the fragment F2-(Ϫ746/Ϫ513) or fragment C-(Ϫ581/Ϫ513) possessed relatively high promoter activities, the activity was masked when a downstream region (Ϫ513/ ϩ183) was included (23). The strong reduction of the transcriptional activity could be due to either a distance effect or to the presence of a GC/GT box acting as a negative element in the downstream region (23). To distinguish between these two possibilities, the fragment F2 or C was constructed into a pCAT-P vector containing the SV40 promoter at a remote position, nearly 2.8 kilobase pairs upstream of the transcriptional start site of the CAT gene. As shown in Fig. 1B, very little or no activation was observed using either of the two constructs, F2-(Ϫ581/Ϫ513)-P or C-(Ϫ581/Ϫ513)-P. These results suggested that the fragment F2 or C functioned in a distance-dependent manner. Taken together, these results suggested that the 69-bp sequence was mainly responsible for the transcriptional activity of the 234-bp fragment and that it functioned in an orientation-independent but a distance-dependent manner.
The 69-bp Sequence Contains Two Functional cis-Elements, a GC box and an E box (CACGTG)-Inspection of the sequence of the 69-bp region revealed a potential Myb site, a GC box with one mismatch, and two types of E box, CAGCTG (class A) and CACGTG (class B). To identify functional cis-elements in the 69-bp region, DNase I footprinting analysis was performed using nuclear extracts from Caco-2 and HeLa cells. Two protected regions, CI and CII, were revealed using both cellular nuclear extracts ( Fig. 2A). Region CI-(Ϫ578/Ϫ559) included a Myb-binding site and a previously described (23) GC box, and region CII-(Ϫ532/Ϫ519) contained a 14-bp palindromic sequence with an E box (CACGTG) of class B in the center of the palindrome (Fig. 2B).
Nuclear proteins binding to the two protected regions were then examined by EMSA using nuclear extracts from Caco-2 cells and the probes covering the two regions, respectively. Two protein-DNA complexes were revealed when DS oligonucleotide CI was used as a probe (Fig. 3A, lane 2). Both complexes were eliminated when competition was carried out using unlabeled oligonucleotide Sp1 containing the consensus sequence of the GC box (lane 3) or the wild-type oligonucleotide CI sequence (lane 4). The formation of the two complexes was not affected when the oligonucleotide CI-M which contained point mutations in the GC box (Table II) or the oligonucleotide Myb was used as competitors ( lanes 5 and 6). These results indi- cated that the binding of the nuclear proteins in region CI occurred specifically at the GC box rather than at the Myb site. Supershift analysis using antibody against Sp1 and Sp3 has indicated that the complex I was formed by interaction with both Sp1 and Sp3, and the complex II was formed by interaction with Sp3 alone (23). Gel shift analysis using oligonucleotide CII as a probe also showed the formation of two complexes (Fig. 3B). However, one of the complexes was very dominant whereas the other was very faint (lane 2). Both complexes were abolished when competed with excess of unlabeled wild-type oligonucleotide CII (lane 3) but not with oligonucleotide CII-M (lane 4) in which the 14-bp palindromic sequence was mutated including the core sequence of the E box (Table II).
Both the GC Box and the E Box Are Required for the Transcriptional Activity of the 69-bp Promoter Region-To evaluate the role of the GC box and the E box on the promoter activity of the 69-bp region, the same mutations that eliminated nuclear protein binding (Fig. 3) were introduced into the plasmid construct C-(Ϫ581/Ϫ513). Fig. 4 shows the CAT activity of the wild-type or the mutant constructs transfected in Caco-2 cells. Mutations in the GC box resulted in about 50% loss of the CAT activity of the 69-bp fragment, whereas mutations in the E box alone or in both the GC box and the E box diminished the transcriptional activity by about 90%. These results indicated that both the GC box and the E box were required for the full transcriptional activation of the 69-bp fragment. However, the relative contributions of these two cis-elements were different. The presence of the E box was essential for the transcriptional activity contributed by the GC box but not the other way around. On the other hand, the presence of the GC box enhanced the transcriptional activity of the E box by 2-fold, suggesting that there could be functional physical interactions between the nuclear proteins that bind to the two sites. In order to test this possibility, gel-shift analysis was carried our using the 69-bp fragment containing both the GC box and the E box.
Nuclear Proteins Bound at the GC Box and the E Box Interact Physically-When the labeled 69-bp promoter fragment was allowed to bind to the nuclear extracts from Caco-2 cells, at least five DNA-protein complexes were revealed (Fig. 5, lane 1). When competed with excess unlabeled oligonucleotide CI (lane 3) or Sp1 (lane 6) which contained the GC box, the complexes I, II, III, and IV were eliminated. When the competition was carried out using unlabeled oligonucleotide CII carrying the E box, the complexes I, II, and V disappeared (lane 4). Competition with oligonucleotide CI-M containing mutations in the GC box (lane 2) or CII-M containing mutations in the E box (lane 5) did not affect the formation of the complexes I-V. These results suggested that the formation of the complexes I and II was the result of physical interactions between nuclear proteins bound to both the GC box and the E box since both the complexes were unable to form in the presence of either unlabeled oligonucleotide CI (lane 3) or CII (lane 4).
Since the 69-bp fragment containing the wild-type E box was essential for driving transcription in Caco-2 cells (Fig. 4), additional studies were carried out using a 29-bp fragment, which corresponded to the footprinting region CII containing the E box, to test whether it is able to drive transcription not only in Caco-2 cells but in other cells as well.
The Oligonucleotide CII (29-bp) Is Able to Direct Bidirectional Transcription in Three Types of Cells-Initially we tested by EMSA whether the 29-bp fragment is able to bind to nuclear factors in K-562 and HeLa cells. As shown in Fig. 6, two DNA-protein complexes were formed when the 32 P-CII fragment was incubated with nuclear extracts from either HeLa (Fig. 6A, lane 1) or K-562 (Fig. 6B, lane 1) cells. This pattern was similar to that obtained using nuclear extracts from Caco-2 cells (Fig. 3B). The formation of the two complexes were eliminated when competed with excess unlabeled oligonucleotide CII (Fig. 6, A and B, lanes 2-4). In contrast, the oligonucleotide CII-M was unable to compete for the formation of the two complexes (Fig. 6, A and B, lanes 5-7). These results suggested that nature of the nuclear proteins binding to the 29-bp was similar or identical in all the three cell lines tested and that the binding was specific to the E box.
The ability of the 29-bp fragment to drive transcription was then tested in Caco-2, HeLa, and K-562 cells. As shown in Table III, the 29-bp fragment was able to drive CAT gene transcription in both orientations in all the three cell lines tested (Table III), and the CAT activity noted, 8 -11-fold, was similar to that of the SV40 promoter. Since the E box-(CACGTG) can interact with multiple transcription factors of the basic helix-loop-helix/leucine zipper (bHLH/LZ) family, the identity of the transcription factor binding to the E box motif was further investigated.
USF Binds to the E box-At least nine members of the bHLH/LZ family of transcription factors bind either as heteroor homodimers to the E box core sequence (CACGTG). Whereas the c-Myc, Max, Mad, and Mxil are involved in the Myc network in controlling cell proliferation and differentiation (32), the others, TFEB, TFE3, and TFEC belong to the Mit subfamily (33) and are involved in the regulation of immunoglobulin heavy chain and insulin gene expression, and USF which is ubiquitously expressed (4) stimulates transcription of many genes (15)(16)(17)(18)(19)(20)(21)(22). Thus, we chose USF and c-Myc as the first candidates to examine their binding to the E box.
As shown in Fig. 7 similar EMSA pattern was obtained using nuclear extract from both Caco-2 (Fig. 7A) and HeLa (Fig. 7B) cells. When the nuclear extracts were incubated with anti-USF1 antibody (Fig. 7, A and B, lane 2), the complex I was  (Table II) containing a GC box with one nucleotide mismatch and a potential Myb-binding site was incubated with (lanes 2-6) or without (lane 1) nuclear extracts from Caco-2 cells. The nuclear extracts were preincubated with binding buffer alone (lane 2) or with unlabeled DS oligonucleotides as indicated (lanes 3-6). B, 32 P-labeled DS oligonucleotide CII (Table II) containing a 14-bp palindrome sequence with an E box in its center was incubated with (lanes 2-5) or without (lane 1) Caco-2 cell nuclear extracts. The nuclear extracts were preincubated with buffer alone (lane 2) or with unlabeled DS oligonucleotide competitors (lanes 2-4) as indicated.
diminished and the complex II was completely abolished, indicating that both complexes contained USF1. When the nuclear extract was incubated with anti-USF2 antibody (Fig. 7, A and  B, lane 3), formation of the complex I was mainly interrupted. This result indicated that complex I represented a combination of both USF1 and USF2. Addition of both anti-USF1 and anti-USF2 antibodies to the nuclear extracts resulted in complete abolishment of the complex I and II (Fig. 7, A and B, lane 4). In contrast, antibody against c-Myc did not affect the formation of the two complexes (Fig. 7, A and B, lane 5). DISCUSSION The present study has provided some insights into the basal and activator-dependent transcription using a human TC II promoter fragment that lacked both a TATA box and an Inr element. By deletion and transient transfection studies, we have demonstrated that a 69-bp DNA sequence from the human TC II promoter possesses bidirectional promoter activity but not an enhancer activity (Fig. 1B, fragment C). The 69-bp fragment did not contain an obvious TATA box or a known Inr element but contained a GC box with one mismatch and a 14-bp palindromic sequence (TGCTCACGTGACCA) with an E box (underlined) in its center. These observations raised an interesting issue as to how the 69-bp DNA fragment without a TATA box and Inr element functioned as a promoter.
Our EMSA experiment (Fig. 3) has shown that both the GC box and the E box were functional in binding to nuclear factors, and immunosupershift analysis further demonstrated that the GC box interacted with Sp1 and Sp3 (Fig. 3, Ref. 23), whereas the E box was recognized by both USF1 and USF2 (Fig. 7). Site-directed mutagenesis (Fig. 4) demonstrated that both the GC and the E box were required for the full promoter activity of the 69-bp fragment. However, their individual contribution toward the promoter activity was not equivalent. Mutations in the GC box reduced the promoter activity by 50%, whereas mutations in the E box alone or in both the E box and the GC box resulted in about 90% reduction of the promoter activity. These results implied that the E box was required for both the GC-dependent and -independent promoter activity of the 69-bp fragment, and there was potential cooperative interactions between nuclear factors that bound to these two cis-elements. Direct evidence for physical interactions between Sp1 or Sp3 and USFs was provided by EMSA (Fig. 5). The essential role of the E box in mediating transcription was further demonstrated by transfection of the fusion plasmid containing the DS oligonucleotide II (29-bp) which contained only the E box in three different cell lines (Table III).
The observation that a short (29-bp) DNA fragment containing only one recognizable cis-element, an E box, is sufficient to direct bidirectional transcription efficiently is very interesting. In general, a eukaryotic core promoter contains either a TATA box or an Inr element that are recognized by the TATA-binding protein (TBP), a component of the TFIID complex, and Inrbinding protein, respectively. TBP or Inr-binding protein plays a central role in recruiting basal transcription factors and RNA polymerase II forming a preinitiation complex (PIC). However, there are a few reports that have demonstrated that an activator-binding site alone is sufficient for a minimal promoter activity in transfected cells. These include the glucocorticoid response element (34) and the Ets motif (35)(36)(37). The mechanism by which the 29-bp sequence functions as a minimal promoter is not known. It is possible that USF bound to the E box stabilized the binding of TFIID to a cryptic TATA element through protein-protein interactions. Alternatively or additionally, USF bound to the E box can recruit TFIID and/or other components of the basal transcription machinery to the promoter and facilitate the assembly of the PIC. Several lines of evidence support this possibility. First, it has been reported (38,39) that TBP is capable of binding to a number of sequences that are completely unrelated to the consensus sequence of the TATA element. Second, USF has been shown to be able to interact with TFIID (12,40), increase the rate or stability of TFIID binding (41), and stabilize formation of the PIC (42). Third, USF is also able to interact with other transcription factors, including TBP associated factor TAFII55 (43), transcriptional cofactor PC5 (44), and TFII-I (45,46) which can bind to both the Inr and the E box.
Another interesting aspect of this study is the distinct roles of the E box and the GC box in the transcriptional activity of the 69-bp sequence. Although the presence of the E box was essential for the transcriptional activity due to the GC box, the presence of the GC box was not required for the transcriptional activity due to the E box. This observation implied that USF mainly played a role in transcriptional initiation while the Sp1 stimulated the transcription. This observation is somewhat surprising since it has been shown (47)(48)(49) before that Sp1binding site could direct transcriptional initiation of several TATA-less promoters. Our finding that USF-binding site was more efficient than that of Sp1-binding site in mediating transcriptional initiation could be a general phenomenon or restricted to a specific context of a promoter or cells. Nevertheless, our finding is in agreement with the hypothesis (50) that USF1 may not directly be responsible for transactivation but functions via interactions with other proteins. In this regard, it is interesting to note that USF1 does not only bind to the E box but also to the pyrimidine-rich Inr element (45,46). In addition, ectopically expressed USF1 could stimulate transcription initiation through Inr element (46). Therefore, it is likely that the E box functions similarly to that of the Inr element if it is located near the transcriptional start site. Our finding that both the 69-bp (Fig. 1B) and the 29-bp sequence (data not shown) mediated transcription in a position-dependent manner also supports a recruiting rather than an activating function of the USF bound to the E box. Indeed, USF-binding site near the transcriptional start site has been found being essential for the basal promoter activity of some TATA-less promoters (15,18,19).
The functional significance of the role of the E box in TC II transcription is not known. Although the binding of USF to the E box may not affect the basal transcription of TC II in vivo due to its distal location from the start sites of the TC II gene, the potential binding in vivo of other members of bHLH/LZ family cannot be ruled out. Some of these could include Myc/Max, Mad/Max, and Max/Max. Elevation of plasma TC II levels is noted in a variety of cancers (51), and our hypothesis at the present time is that the E box by binding to Myc family may increase the transcription of the TC II gene which in turn will result in increased secretion of TC II to the circulation. Future studies with ectopic expression of the proteins belonging to the Myc family will address the role of these transcription factors in the in vivo transcription of the TC II gene.
In summary, we have identified a 69-bp DNA fragment containing a GC box and E box that is able to mediate transcription in vivo in an orientation-independent manner. The transcriptional activity was due to interplay between Sp1/Sp3 and USF1/USF2 that bind to these two cis-elements. Moreover, 29-bp fragment localized to the 3Ј-end of the 69-bp fragment containing only the E box was sufficient by itself to drive transcription, implying that USF bound to an E box can initiate and activate transcription in the absence of TATA box or other known Inr elements.