INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Enterococci normally colonize the intestinal tract, but these organisms, particularly Enterococcus faecalis, are also known to cause many clinical infections in humans, including septicemia, bacteremia, urinary tract infections, and 5 to 15% of cases of bacterial endocarditis (15). The existing knowledge of the factors that may influence the ability of enterococci to colonize host tissues, translocate across epithelial barriers, and survive in different host environments is rudimentary, but their increasing resistance to multiple antimicrobial drugs makes the study of pathogenesis of these organisms all the more important (18).

Interactions with host cells and colonization of mucosal surfaces are considered to be primary events in the pathogenesis of many infections (2). The pathogenesis of bacterial endocarditis is believed to begin with bacterial adhesion to the extracellular matrix (ECM) of damaged heart tissue. Bacterial surface adhesins have been suggested to play a major role in adherence and colonization. Staphylococci are known to bind to a large number of proteins present in the host ECM. Molecular characterization and functional characterization have identified a number of proteins, such as a collagen binding protein, Cna (22), fibronectin binding proteins (12, 29), and fibrinogen binding proteins (3, 5), collectively named microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) (21), that mediate binding to ECM proteins. MSCRAMMs typically share some common structural features: (i) a short signal sequence followed by a nonrepetitive region, which in most cases is responsible for binding to ECM proteins; (ii) a repetitive region that exhibits variation among strains; and (iii) a C-terminal domain that includes an LPXTG anchoring motif and a hydrophobic membrane-spanning domain followed by a short tail rich in positively charged amino acids (8, 21).

Our recent work identified a gene in E. faecalis coding for a putative protein designated as Ace that has characteristics similar to those of the collagen binding protein Cna of Staphylococcus aureus (25). The Ace sequence from E. faecalis strain V583 shows a putative N-terminal signal sequence followed by a 335-amino-acid-long A domain. The B domain is composed of 4.4 tandemly repeated 47-residue units of >90% identity. A cell wall-associated domain rich in proline residues that contains the cell wall-anchoring LPXTG consensus sequence and a hydrophobic transmembrane region of 18 amino acids followed by a short cytoplasmic tail represents the carboxy-terminal end of the protein (25). This work also localized the collagen type I (CI) binding property of Ace produced by E. faecalis strain EF1 to the A domain based on biochemical evidence. More recent results, submitted as a companion paper, demonstrate that Ace mediates the 46°C-evoked adherence of strain OG1RF to collagen type IV (CIV) and mouse laminin (LN) (19) in addition to CI (25).

In the current study, we have studied sequence variation in the E. faecalis ace genes. Since most strains of E. faecalis exhibit conditional binding (i.e., after growth at 46°C), we also attempted to detect Ace proteins from bacterial protein preparations made from cultures grown at both 37 and 46°C. Finally, in an effort to find evidence of expression of ace under more physiological conditions than 46°C, we have examined sera from patients with enterococcal infections for the presence of antibodies to Ace.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. The E. faecalis and Enterococcus faecium isolates used in this study were selected, in most cases, arbitrarily from our laboratory collection (obtained over a 20-year period from various locations in the United States, Argentina, Thailand, Lebanon, and Spain); many of them have been well characterized and are known not to be clonally derived (6, 10, 14, 16, 36). These isolates were obtained from wounds, urine, feces, and blood, including endocarditis. E. faecalis strains OG1RF, JH2-2, and V583 have been described previously (11, 17, 27).

Culture conditions. Enterococci were grown in brain heart infusion broth or agar (Difco Laboratories, Detroit, Mich.) at 37°C for routine purposes or at 46°C. Escherichia coli cells were grown in Luria-Bertani (LB) broth or on LB agar with appropriate antibiotics overnight at 37°C. The concentrations of antibiotics used for E. coli were kanamycin at 50 µg/ml and ampicillin at 50 to 100 µg/ml.

General DNA techniques. Routine DNA techniques were performed by standard methods (28). Chromosomal DNA from E. faecalis was isolated according to the previously described method (17). PCR amplifications were performed with a DNA thermal cycler (Perkin-Elmer Corp., Norwalk, Conn.) and synthetic oligonucleotide primers purchased either from Life Technologies (Grand Island, N.Y.) or from Genosys Biotechnologies, Inc. (Woodlands, Tex.).

Antiserum to the Ace A domain of OG1RF. The cloning and expression of the E. faecalis OG1RF ace gene, coding for all 335 amino acids of the Ace A domain, generation of polyclonal serum against this purified recombinant Ace A, and reactions of this serum with OG1RF have been described elsewhere (19) (Table 2).

Western blotting. Protein extracts from E. faecalis cultures grown at 37 and 46°C were prepared by the mutanolysin extraction method as described in the companion paper (19). Mutanolysin extracts from E. faecalis strains were electrophoresed on 4 to 12% NuPAGE Bis-Tris gels (NOVEX, San Diego, Calif.) under reducing conditions in MOPS (3-[N-morpholino]propanesulfonic acid) buffer and transferred to a polyvinylidene difluoride (PVDF) membrane. The presence of Ace protein was detected by incubation with either the anti-Ace A polyclonal antiserum described above or eluted antibodies from human endocarditis serum (antibody I) followed by protein A-horseradish peroxidase conjugate (antibody II) and development with 4-chloronaphthol in the presence of H₂O₂.

Human sera. From our laboratory collection of sera (collected from different medical centers in the United States), four study groups that were grouped based on the diagnosis of infection were selected for analysis. Serum samples known to have antibodies against enterococcal total proteins from previous studies (1, 33, 39, 40) were included. Strains isolated from patients who had donated serum but which were not available to us, and hence could not be identified to the species level in our laboratory, were classified as enterococcal species unknown (ESU). Sera from 21 patients with E. faecalis endocarditis (including some corresponding to strains studied here) and four patients with ESU endocarditis constituted one group. A second group consisted of nine serum samples collected from patients with E. faecalis nonendocarditis infections, such as bacteremia, urosepsis, and osteomyelitis, and three sera obtained from ESU nonendocarditis infections. The third study group consisted of serum samples from six patients with E. faecium endocarditis, one patient with E. faecium urosepsis, and two patients with streptococcal infections. The final group, consisting of 12 sera obtained from hospitalized patients (HPS) with no knowledge of their diagnosis or of any infection, was included as a nonhealthy control group. Available normal human sera (NHS) from our laboratory collection, previously pooled in groups of two to three from a total of 20 healthy volunteers, were used as a healthy control group.

ELISA. An enzyme-linked immunosorbent assay (ELISA) using human sera was performed by a previously described method with some modifications (1). Polystyrene microtiter plates (Dynatech Laboratories, Inc., Alexandria, Va.) were coated with 50 ng of recombinant Ace A protein from OG1RF in 100 µl of phosphate-buffered saline (PBS) and allowed to incubate overnight at 4°C. Wells were washed five times with PBST (PBS with 0.01% Tween 20). After blocking wells with 3% bovine serum albumin (BSA) at 37°C in PBST, wells were washed three times with PBST. Each serum was assayed in duplicate in serial dilutions of 1:16 to 1:2,048 in 1% BSA. Goat anti-human immunoglobulin G (IgG)-peroxidase conjugate was used for detection of human antibodies to Ace. A₄₅₀ was measured following the addition of 3,3',5,5'-tetramethylbenzidine and H₂O₂. Titers were determined after subtracting values from appropriate negative controls. For control sera, the optical density at 450 nm (OD₄₅₀) was measured at each dilution. The sum of the average OD₄₅₀ value and two times the standard deviation was calculated for each dilution and used as the cutoff value for determining serum titers. A one-tailed Student's t test was used to compare Ace A antibody levels among the four groups of subjects.

Enrichment of Ace-specific antibodies by elution and their effect on adherence. Recombinant Ace protein was electrophoresed in 10% NuPAGE Bis-Tris gels (NOVEX), transferred to a PVDF membrane, and incubated with E. faecalis endocarditis serum S0032. Ace A-specific antibody elution was performed by the procedure described elsewhere (40). Inhibition of enterococcal adherence to CI, CIV, and LN with IgGs affinity purified from normal human sera or from an E. faecalis endocarditis patient serum, S0032, was carried out as described elsewhere for rabbit sera (19). Results are presented as percentage of cells bound, based on the formula (radioactivity of bound cells/radioactivity of total cells added) × 100.

Nucleotide sequence accession numbers. The Ace nucleotide sequences reported here were submitted to GenBank under accession numbers AF260872 to AF260896.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

E. faecalis ace sequences. DNA sequencing and analysis revealed that the ace gene of E. faecalis OG1RF is 2,166 bp in length, encoding a putative polypeptide of 721 amino acids (Fig. 1A). As was previously reported for Ace of E. faecalis strain EF1, the first 31 residues have the properties of a signal peptide, with a charged region followed by hydrophobic residues (25). The N-terminal region is composed of a 335-amino-acid A domain, followed by a tandemly repeated B domain (Fig. 1A). In the B domain, 47 amino acids are repeated five times preceded by a short 20-amino-acid partial repeat (Fig. 1D). Recer (recombinant sites in genes that also serve as flexible spacers in the protein) sequences previously described by de Chateau and Bjork (7), GAA AAT CcA GAT GAA coding for presumably unstructured ENPDE, were identified in the nucleotide sequence at the boundary between each B repeat. The C-terminal region is composed of a cell wall domain with conserved LPKTG anchorage residues, followed by an 18-amino-acid hydrophobic membrane-spanning domain and a short cytoplasmic tail as previously found for EF1 (25). The predicted molecular mass of the Ace protein of OG1RF after signal peptide processing is 75.6 kDa.

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Structural organization of Ace and its variation in different E. faecalis strains. (A) Schematic representation of E. faecalis OG1RF Ace. S, 31-amino-acid putative signal peptide; A domain, 335-amino-acid nonrepetitive binding domain; B domain (5.4 repeats), 20-amino-acid partial repeat followed by five 47-amino-acid repeats separated by recer sequences (GAA AAT CcA GAT GAA coding for presumably unstructured ENPDE); W, cell wall domain; M, membrane-spanning domain; C, charged C terminal. (B) Diagrammatic representation of Ace B domain variants. (C) Variations in Ace A identified in 26 E. faecalis strains collected worldwide. The shaded region represents amino acids 174 to 319 of the E. faecalis Ace protein that corresponds to S. aureus Cna amino acids 151 to 318 known to be critical for collagen binding. XY denotes the respective amino acid change. The number in parentheses denotes the number of strains in which the amino acid change was observed in the 26 sequenced strains. (D) Amino acid sequence of B repeats of OG1RF. Nonidentical amino acids are shaded.

Correlation of in vitro expression of Ace and of adherence. Table 2 summarizes the adherence characteristics and results of probing mutanolysin-phenylmethylsulfonyl fluoride (PMSF) extract concentrates of eight different strains of E. faecalis (as well as results with OG1RF described in the companion paper [19]) with polyclonal immune serum raised against recombinant OG1RF Ace A. After growth at 37°C, a single ~105-kDa protein band was seen in extracts of the E. faecalis END6 strain, and there was a single ~86-kDa weakly positive band for the E. faecalis MC02152 strain (Fig. 2), whereas no band was detected in extracts of the remaining six strains tested. Probing of mutanolysin extracts prepared from these eight E. faecalis strains grown at 46°C with anti-Ace A antibodies showed a single reactive protein band in six E. faecalis strains (Table 2 and Fig. 2). The four observed sizes of protein bands are in concordance with the different numbers of B repeats (Table 2). No band was detected in extracts prepared from LBJ-1 grown at 37 or 46°C, and, as anticipated from sequencing data, no protein band was detected in E. faecalis SE47b. The adherence phenotype of these E. faecalis strains to CI, CIV, and LN was retested, and the results are presented in Table 2. In addition to two previously reported E. faecalis strains, END6 and SE47b, which showed adherence to collagens and/or LN even after growth at 37°C (38), E. faecalis MC02152 grown at 37°C showed low-level binding to ECM proteins (6% to CI, 8.9% to CIV, and 7.1% to LN), while the remaining E. faecalis strains showed <5% binding after growth at 37°C; the latter strains were considered as adherence negative, since we use 5% of cells bound as a cutoff to define adherence. Seven of these strains, excluding LBJ-1, showed a marked increase (to 20%) in adherence to CI, CIV, and LN after growth at 46°C. Of note, strain SE47b, which showed significant binding to CI, CIV, and LN after growth at both 37 and 46°C (38; this study) also showed a high degree of clumping under in vitro culture conditions, which may have resulted in high counts of clumped cells, leading to a high percent of binding by a non-Ace-mediated mechanism at both 37 and 46°C. IgGs purified from anti-Ace A rabbit immune serum were unable to inhibit adherence of SE47b (data not shown).

View larger version (60K): [in this window] [in a new window]   FIG. 2.   Western blot of mutanolysin surface preparations from 37- and 46°C-grown E. faecalis isolates probed with anti-Ace A polyclonal immune rabbit serum. Lanes: 1 and 2, protein extracts from 37- and 46°C-grown MC02152; 3 and 4, protein extracts from 37- and 46°C-grown END6; 5 and 6, protein extracts from 37- and 46°C-grown V583; 7 and 8, protein extracts from 37- and 46°C-grown SE47b; and 9, molecular mass standards.

Reactivity of serum from humans with enterococcal infections with Ace A recombinant protein. We initially screened several E. faecalis endocarditis sera by Western blotting. Among five sera, one (S0032) showed strong reactivity, and three reacted moderately to recombinant Ace A protein, suggesting that in vivo expression of ace by different strains had occurred in these patients (Fig. 3). Serum from a patient with E. faecium endocarditis did not react with recombinant Ace A. 

View larger version (63K): [in this window] [in a new window]   FIG. 3.   Immunoblot of recombinant Ace A protein of E. faecalis OG1RF after probing with sera obtained from patients diagnosed with enterococcal infections. Lanes: 1, molecular mass standards; 2 to 6, sera from different patients with E. faecalis endocarditis; 7, serum from patient with E. faecium endocarditis; and 8, NHS.

View larger version (17K): [in this window] [in a new window]   FIG. 4.   Distribution of anti-Ace A IgG titers in human sera. Efs endo, sera from patients with E. faecalis endocarditis; ESU endo, sera from patients with endocarditis due to ESU; Efs other, sera from patients with E. faecalis nonendocarditis infections; ESU other, sera from patients with ESU nonendocarditis infections; Efm, sera from six patients with E. faecium endocarditis and one patient with E. faecium urosepsis; Strep, patient sera from streptococcal infections; and HPS, sera from hospitalized patients with no knowledge of their diagnosis or of any infection. ESU represents species identified at the time of diagnosis and the strains isolated from patients who had donated serum samples that were not available to us and hence were not identified to the species level in our laboratory.

Ability of IgGs from endocarditis serum to inhibit adherence of E. faecalis OG1RF to ECM proteins. We examined the ability of IgGs purified from a high-Ace A-titer E. faecalis endocarditis patient serum, S0032 (HTS), to inhibit adherence of 46°C-grown E. faecalis OG1RF to CI, CIV, and LN. Preincubation of OG1RF with IgGs from this serum at concentrations greater than 2 mg/ml inhibited adherence to CI, CIV, and LN by about 16- to 24-fold relative to NHS, as shown in Table 3. Purified IgGs from NHS had a negligible effect on adherence at these concentrations.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Inhibition of adherence of E. faecalis OG1RF to ECM proteins by Ace A-specific antibodies eluted from E. faecalis endocarditis patient serum S0032. 35S-labeled bacteria were incubated with 10 µg of eluted Ace A-specific antibodies per ml for 1 h at 37°C. Adherence was tested in wells coated with 1 µg of ECM proteins (see text). Bars represent the percentage of cells bound (mean ± standard deviation) for four wells. Results are representative of two independent experiments.

Lack of evidence of an ace homolog in non-E. faecalis species. Our recent hybridization results with several enterococcal strains with the 1,090-bp ace probe (amplified with AceF2a and AceR3a) indicated that ace is specific to E. faecalis strains (R. W. Duh, K. V. Singh, K. Malathum, and B. E. Murray, submitted for publication). Southern hybridization of DNA preparations from nine E. faecium strains with the 419-bp conserved ace DNA probe under low-stringency conditions showed no bands, further implying the absence of a close ace homolog in E. faecium.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Our earlier investigation has reported a conditional adherence phenotype among most E. faecalis isolates (38). Following this, we identified an E. faecalis gene, ace, that encodes a putative adhesin (Ace) and presented evidence for its role in binding to CI (25). In our companion paper, we disrupted the ace gene in the laboratory strain OG1RF and reported that Ace mediates adherence to CIV and LN in addition to CI (19).

In the current study, we examined the diversity of the ace gene in different E. faecalis strains. Our initial amplification of ace sequences from 15 E. faecalis isolates by PCR showed DNA fragments of four different sizes. To explain this observed size difference and to investigate the extent of differences in ace sequences among E. faecalis isolates obtained from different sources, we sequenced the complete ace gene from eight selected E. faecalis strains. Comparison of the nucleotide and deduced amino acid sequences of Ace from these strains with that available for the V583 strain from The Institute of Genomic Research database showed the highly conserved N-terminal regions representing the A domain, followed by the variable B repeat region. Analysis of these sequences revealed that ace occurred in four different forms relating to variation in the B repeat numbers. Similarly, four molecular sizes of Ace proteins were observed on Western blots probed with anti-Ace A immune rabbit serum. As reported earlier for Ace proteins from E. faecalis strains OG1RF (19), EF1, and EF2 (25), the observed molecular sizes of Ace detected on Western blots of extracts from different E. faecalis strains were found to be larger than the predicted sizes based on deduced amino acid sequences, perhaps due to their highly acidic nature, as shown in Table 2. Consensus 15-nucleotide recer sequences were identified between each B repeat. Earlier analysis of recer sequences in Peptostreptococcus magnus suggested their possible role in recombination of new incoming modules at the DNA level (7). Similarly, at the protein level, the proline residues in ENPDE recer sequences have been proposed to promote lack of structure, thus allowing interdomain flexibility. No recer sequences were reported in staphylococcal collagen binding gene cna. Although we do not have any direct evidence of recombination occurring at recer sequences, this may possibly explain the variation in B repeats. We have yet to characterize the function of the B domain. Although several functions were predicted for the B domains of Cna of S. aureus, recent detailed studies were unable to prove any such functions (9, 24, 32). Further sequencing of the N-terminal ace region that codes for the A domain, the region we previously showed is involved in binding to CI (25), from 17 additional strains collected worldwide showed 97.5% identity, indicating the highly conserved nature of this functional domain. In one of these strains, E. faecalis SE47b, the ace gene was interrupted by a stop codon, as will be discussed further below.

We also attempted to correlate the in vitro production of Ace with the observed phenotype (i.e., binding to ECM proteins CI, CIV, and LN after growth at 37 or 46°C). In Western blots, Ace was detected in extracts of only two E. faecalis strains after growth at 37°C, of which one strain, END6, had been previously noted to bind to CI and CIV after growth at 37°C (38). The other strain, MC02152, which showed a faintly positive band after growth at 37°C, exhibited low-level binding to CI, CIV, and LN. This is in contrast to the majority of E. faecalis strains (for which no band was detected after growth at 37°C) which showed <5% binding after growth at 37°C; since we use 5% of cells bound as a cutoff to define adherence, these isolates were considered as adherence negative. Consistent with the observed binding of 46°C-grown E. faecalis strains to CI, CIV, and LN, the Ace protein was detected in most 46°C-grown E. faecalis strains. With MC02152, a much more strongly positive band was observed on the Western blot after growth at 46°C, and its binding increased to 29% to CI, 38% to CIV, and 41% to LN. Our companion paper also reports identification of a single ~105-kDa Ace protein band from 46°C-grown E. faecalis OG1RF extracts, but not from 37°C-grown extracts (19). With E. faecalis LBJ-1, we were unable to detect an Ace protein band on the Western blot with extracts prepared from 37- or 46°C-grown cells, and it is the only strain that showed no adherence to CI, CIV, and LN after growth at either temperature. Similarly, as anticipated from sequencing data, no Ace protein band was found in extracts of SE47b, the strain whose binding was not reduced by anti-Ace A IgGs, indicating a non-Ace-mediated adherence; this strain shows a high degree of clumping in broth which may explain its apparent binding to ECMs. Thus, the observed conditional expression of Ace protein correlates with conditional adherence (i.e., after growth at 46°C) of E. faecalis strains (38). Since adhesin genes of other pathogenic bacteria have been shown to be environmentally regulated (13, 20, 37), the absence of in vitro production of Ace at 37°C is not unprecedented.

In an effort to determine if there was evidence of Ace expression under physiological conditions, we analyzed the antibody levels to recombinant Ace A by using a diversified serum collection from patients from different medical centers with various types of infections caused by different strains. Our results showed significantly higher anti-Ace A IgG levels among most sera obtained from E. faecalis endocarditis patients as well as in some sera from other E. faecalis infections. The two E. faecalis endocarditis sera that were nonreactive with Ace had much lower total enterococcal antibody levels. Since we lack information about the time of serum collection relative to the onset of illness, it is possible that these negative sera were drawn early in infection. One of six sera from E. faecium endocarditis patients also showed reactivity to Ace A protein. Since Southern hybridization of genomic DNA isolated from this strain with the ace probes, even under low-stringency conditions, showed an absence of hybridization, these antibodies may be the result of a prior infection with E. faecalis. It is of interest that the endocarditis serum from the patient infected with LBJ-1 had Ace A antibodies (titer, 1:256). As described earlier, this strain showed neither conditional adherence nor in vitro Ace expression by Western blots, but the presence of antibodies suggests that Ace was expressed at the time of infection or, possibly, during some prior infection. These results indicate that Ace is commonly expressed in vivo during infection by different strains. Similar to our findings that suggest Ace is produced in vivo, although usually not at levels detectable by our assays when grown at 37°C in vitro, we have observed other antigens that reacted with sera from patients with enterococcal infections, but not with rabbit polyclonal serum raised against protein extracts from a 37°C-grown E. faecalis endocarditis isolate (39). We have also observed this with the polysaccharide gene cluster of E. faecalis, for which we have evidence of in vivo, but not in vitro, production, except for an unusual mucoid strain, which expresses a polysaccharide antigen at a lower temperature (41).

In the bacterial ECM adherence assay, inhibition was obtained with IgGs from a high-Ace A-titer E. faecalis endocarditis patient serum, S0032, and with Ace A-specific eluted antibodies derived from this serum. The eluted Ace-specific antibodies reacted only with an ~105-kDa band from extracts of OG1RF grown at 46°C (but not 37°C), indicating the specificity of the eluted antibodies. A recent study of the antibody response to fibronectin binding protein A in patients with S. aureus infections detected considerable variation in IgG levels that reacted with the ligand binding repeat domain of FnBpA. However, these antibodies were unable to block fibronectin binding (4).

Our recent study showed that ace is specific to E. faecalis, because none of the non-E. faecalis enterococcal isolates hybridized to an ace probe, and ace is present in all E. faecalis isolates regardless of their clinical source (Duh et al., submitted); this is different from what is seen with the staphylococcal ace homolog, cna (encoding a collagen binding adhesin of S. aureus), which is present in only 38 to 56% of S. aureus strains (26, 31, 34). In an effort to identify homologs of ace in E. faecium, we have carried out Southern hybridizations under low-stringency conditions. The absence of a close ace homolog in E. faecium is in contrast to our identification of homologs in E. faecium of other E. faecalis genes (e.g., efaA) (30) and a polysaccharide gene cluster (unpublished observation) under low-stringency hybridization conditions.

In conclusion, analysis of ace sequences from E. faecalis strains collected from patients worldwide showed that the E. faecalis-specific gene ace occurs in at least four different forms, with 97.5% identity in the region encoding the A domain and more apparent variation in the region coding for the B domain, due to variation in the number of repeats. Conditional (after growth at 46°C) in vitro expression of Ace, detected with polyclonal antibodies to OG1RF-derived recombinant Ace A, correlated with our previously described conditional adherence of these E. faecalis strains to ECM proteins. Identification of Ace-specific antibodies in sera obtained from patients with enterococcal infections, especially patients with E. faecalis endocarditis, indicates that Ace is commonly expressed in vivo during infection in humans, not just at 46°C in vitro. Investigation of a possible role for Ace in pathogenesis and elucidation of whether the ability of these antibodies to block adherence of E. faecalis to ECM proteins has any potential protective effects in vivo will be the subject of our future studies.