Performance and characteristics of Fasciola ESAs fractionated by HAC
The Fasciola ESAs are complex mixtures of antigens originated from adult flukes when maintained in vitro in a culture medium for a limited period (typically, 2–24 h12, 13, 18, 26–29). Most Fasciola ESAs are secreted by the parasites into their blind-ending gut and then regurgitated through the oral suckle during an intermittent food in/waste out cycle30, 31. However, structural antigens released from the tegument and/or from the gut may also contribute to the diversity of antigens contained in ESAs.
When specific, the use of combined ESAs may be more reliable to diagnose natural Fasciola infections than single antigens32. In an effort for searching simple analytical and/or semi-preparative methods to increase the specificity of whole Fasciola ESAs, in this study we investigate the usefulness of HAC to fractionate these antigens. The main strategy was to obtain a L-cathepsin rich fraction devoid of most non-specific antigens, as well as other contaminants like heme aggregates, either free or bonded to the heme-binding protein FhHDM-1/MF6p, which may be present in relevant amounts in the Fasciola ESAs preparations9, 10.
HAC is broadly used in preparative biochemistry in the purification of monoclonal antibodies, proteins, and nucleic acids33, 34. The main functional groups of HA are positively charged calcium pairs (C-sites) and negatively charged oxygen atoms associated with phosphate triplets (P-sites)33. HA columns are operated near to neutral pH, typically at the pH range of 6-6.8, a condition at which the column can be regarded as negatively charged and the particles are relatively stable35. To our knowledge, HAC was never used to purify L-cathepsins either from Fasciola or from other sources. However, HAC was previously used to partially purify human cathepsin H from human liver36. In this case, the sample to be purified was applied to the HA column in 20 mM phosphate buffer, pH 6.0, and then eluted with an increasing gradient of the same buffer (150 mM final concentration). In contrast, in our study we used HAC for the negative selection of Fasciola L-cathepsins contained in ESAs using PBS at pH 7.2, i.e, the same buffer previously used to dialyse Fasciola ESAs before storage.
As we indicated in the previous section, two main fractions (HAC-NR and HAC-RET -eluted with 0.5 M sodium phosphate-) were initially collected. In Fig. 1A it can be observed that the brown pigment (heme-containing fraction), typically present in Fasciola ESAs, is trapped at the top of the column (arrow). Complementarily, in Fig. 1B we show a typical chromatogram of Fasciola ESAs fractionated by SEC using a Superdex 75 HR column. As reported previously18, the heme-containing fraction is mainly concentrated in peak II while L-cathepsins eluted in peak IV. The proteins corresponding to peak IV were used for comparative studies with HAC fractions in the present study (see below).
To evaluate the performance of HAC to isolate Fasciola L-cathepsins, we first measured the amount of whole protein and L-cathepsins recovered from HAC-NR and HAC-RET fractions (see Fig. 2). From the starting protein (13.8 mg), a total of 5.7 mg (41.4%) and 4.9 mg (35.5%) were recovered, respectively, from HAC-NR and HAC-RET fractions. Then, after AS precipitation of the proteins in fraction HAC-NR and subsequent filtering (fraction Fi-SOLE; see Fig. 2) a total of 4.1 mg of protein were recovered (29.7%). In contrast, the amount of protein which passed the Millex-GW filter before protein solubilization was negligible (< 5% of the incoming protein; see Fig. 2, fraction Fi-NR). It also should be noted that the percentage of protein in Fi-SOLE was only slightly lower than that obtained from peak IV after SEC of Fasciola ESAs using the Superdex 75 HR column (36.9% in average), also measured by the BCA method. However, when tested by the MM3-COPRO ELISA, which specifically detects L-cathepsins, the proportion of Fasciola L-cathepsins present in each of these fractions was very different (Fig. 3). While 96.3% of recovered L-cathepsins were present in the Fi-SOLE fraction, only 3.7% could be found in fraction HAC-RET. These percentages were extrapolated from the data of Fig. 3, considering i) the dilution of samples required to reach the limit of detection of the assay with the test MM3-COPRO (i.e., 1/4096 for Fi-SOLE and 1/128 for HAC-RET) and ii) that the starting dilutions for all samples were adjusted at the same concentration (40 ng/mL, measured by BCA). When the proteins from a second batch of F. hepatica ESAs were chromatographed using the same procedure, the percentages of L-cathepsins we obtained from fractions Fi-SOLE and HAC-RET were similar, which indicates that the proposed HAC method is reproducible. Nevertheless, as no method exists that measures with precision the amount of protein when a mixture of proteins is present in an aqueous sample, and the proportion of protein species in ESAs is variable, the calculated amount of protein in each fraction may be different. For example, the total amount of protein recovered in the final fraction Fi-SOLE dropped from 4.1 mg (measured by BCA) to 2.3 mg when the protein concentration was measured as OD280/2.4, i.e., considering the extinction coefficient and MW of a F. hepatica L1-recombinant cathepsin (see previous section). In any case, from the above results we concluded that: i) only a minimal amount of L-cathepsins and/or their fragments present in Fasciola ESAs (< 5%) was retained by the HA column under the chromatographic conditions used in this study, ii) most of the cathepsins in ESAs can be recovered in the Fi-SOLE fraction, iii) the detection limit of the MM3-COPRO22 drops from 150 pg/mL, when whole ESAs are used, to about 20 pg/mL, when an enriched L-cathepsin fraction (e.g., Fi-SOLE) is used to construct the calibration curve.
Regarding the ability of the HA column to retain heme and heme complexes with protein FhHDM-1/MF6p9, 37, we carried out an UV/VIS absorption spectrum (250–650 nm scan) with each of the fractions obtained by HAC. For comparisons, we also carried out spectra of whole Fasciola ESAs, and the fractions corresponding to peaks I, II and IV obtained by SEC on the Superdex 75 HR column (Fig. 1B) from whole Fasciola ESAs. As can be seen in Fig. 4, with respect to protein content (λmax 280 nm), a high proportion of heme (λmax 405 nm) was present in fractions corresponding to peak II, as previously reported18, and in HAC-RET. However, a small, but still significant, amount of heme was also observed in peak IV. This probably corresponded to small FhHDM-1/MF6p complexes with MW near that of L-cathepsins that cannot be sieved by SEC. In contrast, the HAC-NR had negligible amount of heme, which is indicative of a very high binding affinity of heme to the HA matrix, which is according to the fact that most of the brownish pigment present in Fasciola ESAs remained at the top of the HA column during the entire chromatographic process (see Fig. 1A), much of which persisted bond to HA after the final wash with 1 M NaOH before storage (not showed). It is important to note that the HA column containing the Bio-Gel HTP particles can be normally regenerated using 0.5 M phosphate buffer, according to the manufacturer's instructions. However, complete removal of strong binders/insoluble substances (as in the case of heme and heme complexes present in ESAs) could not achieved even after washing with 1–2 M NaOH. So, removal of the top layer of the HA bed, or discarding the HA matrix when saturated may be necessary.
Analysis of the antigenicity of Fasciola ESAs fractions isolated by HAC
Once we demonstrated that most of Fasciola L-cathepsins present in ESAs were collected in the flow-through (HAC-NR fraction) of HAC, we tested the sensitivity and specificity of such antigens and those remained in fraction Fi-SOLE. For comparisons, we also tested the antigens present in peak IV obtained by SEC, as well as those present in whole ESAs. So, four antigenic preparations (HAC-NR, Fi-SOLE, Peak IV and whole ESAs) were tested in parallel. A total of 130 serum samples from sheep (40 sera from Fasciola-infected and 90 sera from non-infected animals) and 88 from cattle (44 sera from both infected and non-infected animals) were analysed. As no cut-off value was available for these antigens, a ROC analysis was done with each of these antigens. The results of these analyses are showed in Tables 2 (ovine) and 3 (bovine). Except for whole ESAs, the cut-off values obtained were lower for sheep than for cattle. Regarding sensitivity and specificity, whole ESAs showed, as expected, the worst results with values of sensitivity and specificity below 90% for both sheep and cattle. Moreover, the specificity of this antigen only reached a value of 56.8% for bovine serum samples (Table 3). With respect to the HAC-NR antigenic fraction, the values of sensitivity were around 95% for both animal species, but the specificity was better for sheep (98.9%) than for cattle (86.4%). These results suggested that some non-specific antigens are not retained by the HA column and elute together with Fasciola L-cathepsins in the HAC-NR fraction. However, after precipitation of the HAC-NR fraction with 50% AS (fraction Fi-SOLE), the sensitivity and specificity values increased, respectively, up to 100% and 98.9% for sheep, and 97.7% and 97.7% for cattle (Tables 2–3). These values were similar to those obtained with antigens present in peak IV (100% sensitivity and 98,9% specificity for sheep, and 97.7% sensitivity and 100% specificity for cattle), an antigenic mixture that we have already demonstrated to be sensitive and specific enough to be used as target in iELISA for the serodiagnosis of sheep infections caused by F. hepatica18.
Table 2
Results of ROC curve analysis in ovine negative (n = 90) and positive (n = 40) serum samples.
| Cutoff (OD) | Sensitivity (%) [95% CI] | Specificity (%) [95% CI] | AUC [95% CI] | P-value |
ESAs | 0.639 | 87.5 [73.2–95.8] | 85.6 [76.6–92.1] | 0.928 [0.869–0.966] | < 0.001 |
Peak IV | 0.067 | 100 [91.2–100] | 98.9 [94–100] | 1 [0.971-1] | < 0.001 |
HAC-NR | 0.178 | 95 [83.1–99.4] | 97.8 [92.2–99.7] | 0.99 [0.954–0.999] | < 0.001 |
Fi-SOLE | 0.1 | 100 [91.2–100] | 98.9 [94–100] | 1 [0.971-1] | < 0.001 |
Table 3
Results of ROC curve analysis in bovine negative (n = 44) and positive (n = 44) serum samples.
| Cutoff (OD) | Sensitivity (%) [95% CI] | Specificity (%) [95% CI] | AUC [95% CI] | P-value |
ESAs | 0.418 | 84.1 [69.9–93.4] | 56.8 [41-71.7] | 0.789 [0.656–0.844] | < 0.001 |
Peak IV | 0.203 | 97.7 [88-99.9] | 100 [92–100] | 0.997 [0.953-1] | < 0.001 |
HAC-NR | 0.477 | 95.5 [84.5–99.4] | 86.4 [72.6–94.8] | 0.965 [0.903–0.993] | < 0.001 |
Fi-SOLE | 0.303 | 97.7 [88-99.9] | 97.7 [88-99.9] | 0.997 [0.954-1] | < 0.001 |
The data in Figs. 5–6 showed the individual OD values obtained for each method after testing sera from infected (closed circles) and non-infected (open circles) sheep (Fig. 5) and cattle (Fig. 6). As can be observed, the OD values obtained for most sera from infected sheep and cattle were high for all tested antigens. Also, as expected, the observed OD values obtained for sera from non-infected sheep and cattle were much lower for purified antigens than for whole ESAs. In addition, when more conservative cut-off values were selected to ensure 100% specificity (see blue lines in Figs. 5 and 6), the sensitivities obtained for both antigens were also similar: 97.5% (Fi-SOLE and Peak IV) for ovine samples, and 95.5% (Fi-SOLE) and 97.7% (Peak IV), for bovine samples. The sensitivity and specificity values obtained for sheep and cattle with the antigenic fraction Fi-SOLE were like those previously reported for commercial devices using purified Fasciola ESAs38.
Proteomic and immunological analysis of Fasciola ESAs fractions isolated by HAC and SEC
To identify the proteins in HAC-NR, Peak IV and Fi-SOLE fractions, we separated them in an SDS-PAGE gradient gel (8–16%) and the stained bands were scissed and processed for identification by nanoUHPLC-Tims-QTOF MS/MS spectrometry. The results of the SDS-PAGE study are showed in Fig. 7. As can be seen, most of the proteins belonging to fractions HAC-NR (lane 1) and Peak IV (lane 3) are grouped into three broad bands with MWs of 24–30 kDa (rows “a”), 14–17 kDa (rows “b”) and 10–13 kDa (rows “c”). The proteins of fraction Fi-SOLE (lane 4) presented the same pattern except that the band “b” was almost absent (see lane 4b, arrow). The nature and relative abundance of the proteins present in each SDS-PAGE band identified by mass spectrometry were summarized in Table 4 and Supplementary Table S1. In Table 4, the identified proteins were grouped by categories (i.e., protein families) and spectral abundance (the number of spectra used to identify each protein), the latter which was used here as a surrogate of the relative abundance of each protein according to the studies of Liu et al. (2004)39. It was reported that this method may have some shortcomings40, particularly when mass analysis is done under different experimental conditions, and since larger proteins contribute with more peptides than shorter ones. However, as we only compared peptides/proteins of similar molecular mass and all protein digestions and MS/MS analysis were done under the same experimental conditions, it is expected that our data accurately reflected the relative abundance of identified proteins. Single proteins, or protein families identified with a minimal of 15 spectra were annotated by their specific names in Table 4, while the remaining low-abundant proteins were included as a single miscellaneous group (others) within each category.
Table 4
Distribution and relative abundance of proteins identified by MS/MS spectrometry in F. hepatica antigen fractions (HAC-NR, Peak IV and Fi-SOLE) after SDS-PAGE analysis (see Fig. 7). (*) Only protein families identified by 15 or more spectra were individually represented. Individual proteins covering 5% or more of total spectra within each protein band were noted in bold. The total number of analyzed spectra was 4084 (HAC-NR), 3350 (Peak IV) and 814 (Fi-SOLE).
SDS-PAGE MW range | HAC-NR (Lane 1) | PEAK IV (Lane 3) | Fi-SOLE (Lane 4) |
PROTEIN FAMILY* | SPECTRA NUMBER (%) | PROTEIN FAMILY* | SPECTRA NUMBER (%) | PROTEIN FAMILY* | SPECTRA NUMBER (%) |
24–30 kDa (bands “a”) | L-cathepsins Actins Others (n = 15) | 1535 (95.3) 16 (1.0) 59 (3.7) | L-cathepsins GSTs Others (n = 14) | 947 (92.6) 26 (2.5) 50 (4.9) | L-cathepsins Others (n = 1) | 730 (99.9) 1 (0.1) |
14–17 kDa (bands “b”) | L-cathepsins FABPs Myoglobin Haemoglobin HSP-70 Thioredoxin Globin-3 Cystatin SODs Histones Others (n = 20) | 486 (28.3) 447 (26.1) 279 (16.2) 238 (13.9) 44 (2.6) 38 (2.2) 32 (1.9) 20 (1.2) 16 (0.9) 15 (0.9) 100 (5.8) | FABPs Myoglobin Haemoglobin L-cathepsins Globin-3 Others (n = 27) | 665 (50.0) 236 (17.7) 214 (16.1) 72 (5.4) 28 (2.1) 115 (8.6) | Others (n = 3) | 18 (100) |
10–13 kDa (bands “c”) | L-cathepsins Thioredoxin Myoglobin Beta-tubulin FABPs Polyubiquitins Cystatin Haemoglobin Others (n = 20) | 307 (40.4) 94 (12.4) 75 (9.9) 47 (6.2) 42 (5.5) 40 (5.3) 40 (5.3) 34 (4.5) 80 (10.5) | FABPs L-cathepsins Myoglobin Haemoglobin Thioredoxin Ca-binding proteins Kunitz Stefin-1 Others (n = 23) | 301 (30.2) 193 (19.4) 112 (11.2) 96 (9.6) 84 (8.4) 83 (8.3) 20 (2.0) 17 (1.7) 91 (9.1) | L-cathepsins Others (n = 12) | 34 (53.3) 31 (47.7) |
When analysing the nature of proteins present in bands “a” (lanes 1, 3 and 4; Fig. 7), it was observed that most of them belong to the family of L-cathepsins, with relative abundances of 95.3% (HAC-NR), 92.6% (Peak IV) and 99.9% (Fi-SOLE). The protein content in bands “b” (lanes 1 and 3; Fig. 7) were more diverse, although four protein families predominated: L-cathepsin fragments, FABPs, myoglobin and haemoglobin. Among them, L-cathepsin fragments and FABPs were the most abundant proteins in the HAC-NR fraction (54.4%), while FABPs predominate in the Peak IV fraction (50.0%). In the SDS-PAGE region corresponding to lane 4b (Fi-SOLE) only small amounts of L-cathepsin fragments and cystatin were detected, all of them below the threshold of 15 spectra (Supplementary Table S1), although this data should be taken with caution since the whole amount of protein in lane 4 was lower than in lanes 1 and 3. Regarding the proteins in bands “c” (lanes 1, 3 and 4; Fig. 7), fragments of L-cathepsins (40.4%), thioredoxin (12.7%) and myoglobin (9.9%) predominated in HAC-NR, while fragments of FABPs (30.2%), L-cathepsins (19.4%) and myoglobin (11.2%) were the most abundant in Peak IV. Finally, fragments of L-cathepsins (53.3%) were the major protein components in fraction Fi-SOLE.
Comparing the diversity and predominance of proteins identified by MS/MS analysis of Peak IV, HAC-NR, and Fi-SOLE fractions, it would be expected that the Fi-SOLE fraction was the most specific in ELISA since this fraction is the one that contains the greater proportion of cathepsins and fragments thereof. However, at a first glance, neither the abundance of L-cathepsins nor the presence of other accompanying majority proteins could explain why the antigens in Peak-IV fraction are much more specific than those present in fraction HAC-NR (Table 4). So, to shed some light on this apparent contradiction, we first analysed which proteins identified by MS/MS analysis in the HAC-NR fraction (low specificity) were not present in either Peak-IV (highly specific) or Fi-SOLE (highly specific) fractions. The list of unique proteins present in fraction HAC-NR is showed in Table 5. As can be observed, β-tubulin > heat shock protein 70 (HSP-70) > actins > histones were the four most abundant proteins that were absent in Peak IV or Fi-SOLE fractions. These proteins are probably originated from the tegumental coat of the flukes, which is constantly released during the in vitro culture of flukes, and probably in more quantity when cultured for long time as with our ESAs preparation (24 h). Using a similar technology for protein identification (LC-MS/MS), the presence of actin, HSP-70, β-tubulin and other glycoproteins was already reported by Ravida et al.11 in a tegumental extract of F. hepatica obtained by treatment of adult flukes with Nonidet P-40 and further purification by lectin affinity chromatography.
Table 5
Diversity and relative abundance of unique proteins identified by MS/MS spectrometry in F. hepatica antigen in fraction HAC-NR which were not identified in fractions Fi-SOLE or Peak IV. Individual proteins covering 0.5% or more of total HAC-NR spectra (n = 4084) were noted in bold. The spectra number of unique proteins in fraction HAC-NR was n = 156.
UNIQUE PROTEINS | SPECTRA NUMBER | PERCENTAGE * |
Actins [Fasciola hepatica] | 27 | 0.66 |
Beta-tubulin [Fasciola hepatica] | 47 | 1.15 |
Collagen type IV alpha 2 chain [Fasciola hepatica] | 1 | 0.02 |
EGF region [Fasciola hepatica] | 1 | 0.02 |
Fibrillin-2 [Fasciola hepatica] | 1 | 0.02 |
Fructose-biphosphate aldolase B [Fasciola hepatica] | 1 | 0.02 |
Heat shock protein 70 [Fasciola hepatica] | 44 | 1.08 |
Histones [Fasciola hepatica] | 24 | 0.58 |
Nuclear transport factor [Fasciola hepatica] | 1 | 0.02 |
Tegumental CaBP4 [Fasciola hepatica] | 6 | 0.15 |
Toll-interacting protein B [Fasciola hepatica] | 1 | 0.02 |
Twitchin [Fasciola hepatica] | 2 | 0.04 |
Anti-β-tubulin, anti-HSP-70, anti-actin and/or anti-histone antibodies have long been reported in serum of normal individuals41, 42, but their presence is more frequently associated with autoimmune and inflammatory human disorders43–46. Likewise, anti-tubulin47, anti-HSP-7048 and anti-histone49 antibodies were also reported in serum from some animal species, but the available data on this topic are scarce. However, given the low proportion of these proteins in the HAC-NR antigenic fraction compared with the remaining proteins in such fraction (0.58–1.15%, see Table 5) it seems improbable that they were important cross-reactive target antigens responsible for the high background OD values observed with some serum samples obtained from non-infected cattle (see Fig. 6). Nevertheless, as the proteins included in Table 5 were not present in fractions Peak-IV or Fi-SOLE, we cannot completely rule out the possibility that one or more of such proteins could be targeted by antibodies from non-infected animals.
Regardless of proteins included in Table 5, there is also the possibility that antigens in the HAC-NR fraction were contaminated with foreign non-protein antigens (e.g., lipopolysaccharides, LPSs), or that some of them were decorated with ubiquitous cross-reacting epitopes (e.g., PC). LPSs (= endotoxins) constitute a family of structurally related glycolipids which are the major constituents of the outer membrane of most Gram-negative bacteria50. LPSs are frequent contaminants of laboratory preparations, and we have previously observed that Fasciola hepatica ESAs may be contaminated with these molecules10. However, in this study we excluded that LPSs were present in the HAC-NR fraction for three reasons: i) LPSs bind with high affinity to HA in the buffer used to obtain HAC-NR antigens33, 51, ii) the treatment of the HAC-NR fraction with allantoin crystalline powder (a recognized strong endotoxin adsorbent52, 53) did not alter the antigenicity of HAC-NR fraction (data not shown) and iii) LPSs are precipitable by ammonium sulphate54 under similar conditions as that used to obtain the specific antigenic fraction Fi-SOLE.
PC is a small lipid-related hapten composed of a negatively charged phosphate and a positively charged choline group. PC is present in a wide/great variety of bacteria, fungi, protozoa, and helminths55–58. In general, PC can be found covalently attached to N-acetylglucosamine from N-glycans and glycolipids59–61 and is considered both a DAMP (danger-associated molecular pattern) and a PAMP (pattern-associated molecular pattern)62. Early studies by Sloan et. al.63 reported the presence of abundant immunodominant PC-bearing antigens in F. hepatica extracts, but the nature and anatomical location of them was never investigated. As the ELISA results presented above indicate that the specificity of the antigens contained in different fractions obtained from Fasciola hepatica ESAs is different, we investigated the presence of PC-bearing antigens in such fractions. Accordingly, we measured the reactivity of seven two-fold serial dilutions of mAb BH8 (starting dilution 1/5000) against four ESAs fractions (whole ESAs, Peak IV, HAC-NR, Fi-SOLE) and PC-Ova (positive control) in iELISA. The data in Fig. 8A show that PC-bearing antigens were mainly present in whole ESAs followed by fraction HAC-NR, but not in the more specific Peak IV and Fi-SOLE fractions. These results suggest that most of the PC-bearing antigens in whole ESAs remain soluble in the filtrate after precipitation with 50% AS and filtration of the HAC-NR fraction (fraction Fi-NR, see Fig. 2). This hypothesis was later confirmed when we obtained strong reactivity of mAb BH8 with PBS-dialyzed antigens from the Fi-NR fraction in iELISA, but not against the HAC-RET antigens eluted with 0.5 M phosphate, which is recommended by the manufacturer to regenerate the HA column (not shown). However, as the protein concentration measured in Fi-NR was very low (see above), and no sugar traces could be detected using the phenol-sulphuric method20, knowing of the nature of the PC-bearing antigens in such fraction requires additional research.
The data of the IHQ presented in Fig. 8B also showed that, like L-cathepsins24, PC-bearing antigens are mainly present in the cecal epithelium, but also in other Fasciola structures as testes. The location of such antigens in the digestive tract of the parasite explains why PC-bearing antigens can be detected by mAb BH8 in ESAs after in vitro culture of the flukes. From these data, we tentatively suggest that PC-bearing molecules in ESAs may be involved in the cross-reactivity phenomenon observed when testing sera from some non-infected animals if whole ESAs are used as target in ELISA. Consistent with this hypothesis, preliminary ELISA inhibition studies in our laboratory showed that most of the cross-reactivity observed with sera from uninfected cattle can be eliminated by incorporating PC into the dilution buffer. Nevertheless, more studies will be required to definitively confirm this hypothesis and to know the nature, structure, and immunological relevance of the PC-bearing molecules detected in Fasciola hepatica ESAs.
In summary, in this article we demonstrated that negative selection of Fasciola hepatica ESAs by HAC combined with AS precipitation is a rapid, robust, cheap, and simple method to obtain an antigenic fraction rich in secreted native L-cathepsins. Purified antigens by this method can be used to improve the specificity of in-house ELISA methods targeting whole ESAs, but also to obtain antigens intended for the immunization of animals to produce anti-cathepsin L (mixtures of CL1, CL2, CL5) polyclonal antibodies. The proposed method can also be used as a first step in the purification until homogeneity of Fasciola hepatica L-cathepsins is achieved. Finally, our study strongly suggest that PC-bearing molecules are major cross-reactive antigens present in F. hepatica ESAs.