Molecular Detection of the Macrolide Efflux Gene: To Discriminate or Not To Discriminate between mef(A) and mef(E)

The presence of a novel macrolide efflux system in streptococci was first described and firmly established in 1996 by Sutcliffe et al. (81). This system was phenotypically recognized and characterized to confer low-level resistance (MICs, 1 to 32 μg/ml) to 14- and 15-membered macrolides but not to 16-membered macrolides, lincosamides, or streptogramin B (or their analogues). This phenotypic pattern of antibiotic resistance was referred to as M-type resistance and is in contrast to the MLS_B phenotype, which confers constitutive high-level resistance (MICs, ≥256 μg/ml) to macrolides, lincosamides, as well as streptogramin B. In the same year Clancy et al. (16) identified the gene responsible for this novel efflux system in Streptococcus pyogenes, and it was designated mef(A). This gene was deposited in the public DNA databases and could be considered the reference sequence for the mef(A) gene (GenBank accession number U70055). Tait-Kamradt et al. (84) later identified a similar gene in Streptococcus pneumoniae that at the time was designated mef(E). Likewise, this gene could be considered the reference sequence for mef(E) (GenBank accession number U83667).

The coding sequences of these two genes appeared to share 90% identity at the DNA level (Fig. 1). Remarkably, in contrast to what might be expected due to the degeneracy of the genetic code, they share only 88% identity at the protein level (48 mismatches in a protein of 405 amino acids). The encoded proteins are strongly hydrophobic, apparent integral membrane proteins with 12 transmembrane segments (16). Because of the high degree of similarity between the mef(A) and the mef(E) genes, Roberts et al. (74) suggested in a minireview that both genes be referred to as just a single class, mef(A). The result of this recommendation would be that if there would be a need to discriminate between the two genes, the recommended nomenclature would be something like mef(A) subclass mef(A) to indicate mef(A) and mef(A) subclass mef(E) to indicate mef(E). In order to increase the readability, we prefer to use the original names mef(A) and mef(E) throughout this minireview.

FIG. 1.

Alignment of the mef(A) and mef(E) nucleotide sequences and the corresponding amino acid sequences. All sequences were taken from the original database entries [GenBank accession number U70055 for mef(A) and MefA (16); GenBank accession number U83667 for mef(E) and MefE (84)]. Dots represent residues identical to that in the mef(A) nucleotide sequence or the MefA amino acid sequence.

In a number of reports it has since been shown that a number of marked differences between mef(A) and mef(E) exist. For instance, the genetic elements carrying mef(A) or mef(E) and their contexts have been studied by Santagati et al. (75), Gay and Stephens (29), and Del Grosso et al. (23) and were shown not only to be quite different but also to behave quite differently. The two genes have disseminated markedly differently and are being recognized in an ever growing number of microbial species. At present, both the mef(A) and the mef(E) genes have unambiguously been identified in five streptococcal species, whereas mef(E) has been identified in five more streptococcal species and in nine additional nonstreptococcal species (Table 1). Furthermore, Amezaga et al. (5) reported that the MICs for mef(A)-containing S. pneumoniae isolates were significantly higher than those for mef(E)-containing isolates. This indicates that, despite the high degree of homology between the two genes, in the context of the genome in which they are embedded, the differences between them are sufficient to impose different susceptibility characteristics on the strains carrying the genes. The existence of these differences between the two genes has prompted others to suggest that the difference between the two genes be maintained (23, 55) and may have been one reason(s) why others also continued using the names mef(A) and mef(E) after publication of the nomenclature minireview (5, 12, 13, 15, 19, 23, 35, 54, 55, 66).

TABLE 1.

Dissemination of mef genes among microbial species

Organism	mef(A)	mef(E)	Reference(s)
Streptococcus pyogenes	+	+	9, 12, 13, 15, 19
Streptococcus pneumoniae	+	+	5, 9, 23, 54, 55, 62
Streptococcus agalactiae	+	+	10
Streptococcus mitis	+	+	8, 66
Streptococcus oralis	+	+	8, 66
Streptococcus salivarius		+	8
Streptococcus anginosus		+	8, 35
Streptococcus intermedius		+	45
Group C Streptococcus		+	9
Streptococcus sp.		+	45
Enterococcus faecalis		+	45
Enterococcus sp.		+	46
Staphylococcus aureus		+	46
Staphylococcus haemolyticus		+	45
Staphylococcus intermedius		+	45
Staphylococcus sp.		+	45, 46
Neisseria gonorrhoeae		+	18
Granilucatella adiacens		+	Unpublished data
Gemella haemolysans		+	Unpublished data

As a result of this, there is no widespread consensus about the nomenclature for the mef genes in the present literature. For readers unaware of this, this may give rise to conflicting interpretations of the available literature and resources on the subject. In this minireview, we outline how the use of different mef gene nomenclatures has created considerable confusion in the field. We also review the current resources on the subject: we performed a search for mef gene sequences in public DNA databases and looked at methods for the detection of mef genes in clinical isolates and tools that can be used to discriminate between mef(A) and mef(E). This information was then used to review the data in the literature with respect to reported genes versus the actual genes that were studied, given the information in the methods sections.

mef GENE SEQUENCES IN PUBLIC DNA DATABASES

Macrolide efflux (mef) genes have been detected in an ever increasing number of different microbial species. We first investigated whether or not these genes were identical to the genes originally detected in S. pyogenes and S. pneumoniae. For this, a search of the general public DNA databases for the original mef(A) coding sequence (GenBank accession number U70055; Fig. 1) (16) was performed with the BLAST program (http://www.ncbi.nlm.nih.gov/BLAST/). This yielded a total of 24 mef-related sequences. All but two were identified in gram-positive cocci, with the exceptions being Neisseria gonorrhoeae and Bacteroides ovatus. These 24 sequences were deposited in the databases as 9 mef(A) sequences, 5 mef(E) sequences, and 10 sequences identified as macrolide efflux (mef) genes without further specification (Table 2). A DNA sequence alignment of the coding regions was made by using the ClustalX program (4) to identify their similarities to the originally described mef(A) and mef(E) genes. The five deposited mef(E) gene sequences were 100% identical to the originally described mef(E) gene sequence (GenBank accession number U83667) (84). However, various nucleotide differences were found among the sequences that were deposited as mef(A). It turns out that only three mef(A) sequences proved to be 100% identical to the original mef(A) gene sequence. Four of nine sequences from gram-positive organisms deposited as mef(A) sequences actually proved to be 100% identical to the original mef(E) sequence. In addition, the gene from N. gonorrhoeae (GenBank accession number AY319932), deposited as a mef(A) gene, turned out to be a mef(E) gene variant (>99% identical). The ninth mef(A)-like sequence identified in B. ovatus proved to be 90% identical to the original mef(A) gene sequence and 86% identical to the original mef(E) gene sequence. The encoded protein, however, lacks the amino-terminal 29 residues and 66 carboxy-terminal residues found in mef(A) and contains only 8 rather than 12 putative transmembrane segments. Also, no apparent low-level erythromycin resistance could be attributed to this gene (88). Therefore, this particular gene sequence is unlikely to encode an actual macrolide efflux pump. Thus, of the nine mef(A) sequences in the public databases, only three sequences proved to be actual mef(A) gene sequences. Because most of these nine mef(A) sequences were submitted after publication of the minireview by Roberts et al. (74), use of the name mef(A) in the deposition of these sequences would be according to the previously recommended nomenclature. However, this recommendation did not include comments that indicated if the sequence actually represented the original mef(A) or mef(E) sequence, which would have been more appropriate. Both genes were just considered to be a single class, mef(A). Of the 10 gene sequences that were deposited as being a mef gene without further specification, the sequences of 4 genes (identified in Granulicatella adiacens isolates and Gemella haemolysans) were 100% identical to that of mef(E), whereas the sequence of 1 (from a group G Streptococcus) was >99% identical to the original mef(A) sequence. The remaining five mef sequences (all from group G streptococci) represent novel mef genes or variants. Two of these five are both 96% identical to mef(E) and 90% identical to mef(A) (GenBank accession numbers AY355405 and AY355406). The remaining three (GenBank accession numbers AY355408, AY355409, and AY355410) are all 90% identical to mef(E), 88% identical to mef(A), and 90% identical to the two sequences with GenBank accession numbers AY355405 and AY355406. Since these novel mef genes have only recently been published, they fall beyond the scope of this minireview.

TABLE 2.

Overview of mef gene sequences in the public DNA databases^a

GenBank accession no.	Species	Yr of submission	Subjectb	Sequence corresponds tob:	Reference
U70055c	Streptococcus pyogenes	1996	A	A	16
U83667c	Streptococcus pneumoniae	1997	E	E	84
AB011259	Streptococcus pneumoniae	1998	E	E	85
AF227520	Streptococcus pneumoniae	2000	A	A	75
AF227521	Streptococcus pyogenes	2000	A	A	75
AF274302	Streptococcus pneumoniae	2000	E	E	29
AF376746	Streptococcus pneumoniae	2001	E	E	23
AJ318993	Streptococcus salivarius	2001	E	E	79
AY064721	Staphylococcus aureus	2001	A	E	46
AY064722	Streptococcus intermedius	2001	A	E	46
AY071835	Enterococcus sp.	2001	A	E	46
AY071836	Streptococcus sp.	2001	A	E	46
AJ557257	Bacteroides ovatus	2003	A	mef(?)	88
AY319932	Neisseria gonorrhoeae	2003	A	E	18
AY355405	Group G Streptococcus	2003	mef	mef	92
AY355406	Group G Streptococcus	2003	mef	mef	92
AY355407	Group G Streptococcus	2003	mef	A	92
AY355408	Group G Streptococcus	2003	mef	mef	92
AY355409	Group G Streptococcus	2003	mef	mef	92
AY355410	Group G Streptococcus	2003	mef	mef	92
AY422726	Granulicatella adiacens	2003	mef	E	Unpublished data
AY422727	Granulicatella adiacens	2003	mef	E	Unpublished data
AY422728	Granulicatella adiacens	2003	mef	E	Unpublished data
AY422729	Gemella haemolysans	2003	mef	E	Unpublished data

^aSee text for further details.

^bA, mef(A); E, mef(E).

^cGenBank accession numbers U70055 and U83667 can be considered reference sequences for mef(A) and mef(E), respectively.

Because only three of nine mef(A) sequences in the public DNA databases proved to be actual mef(A) gene sequences, we conclude that the use of public databases as a sole resource for mef(A) gene sequences may result in the use of improper reference sequences. This may have contributed to the current confusion about the mef gene nomenclature.

DNA-BASED DETECTION OF THE mef GENES IN CLINICAL ISOLATES

Several techniques for the detection of the mef genes in clinical isolates have been described. However, in the large majority of the studies in the literature examined (73 of 77 articles, excluding fundamental publications on the subject), detection of mef gene sequences by molecular biology-based methods is performed by PCR as either the primary or secondary screening procedure. In the remaining studies, some form of DNA hybridization assay was performed (for instance, dot blotting or microwell hybridization assays).

It is not much of a surprise to find that PCR is by far the most established method for the detection of mef genes. No less than 14 different PCR primer combinations for amplification of the mef gene have been reported up to now, and these create amplification products ranging from 202 to 1,759 bp (Table 3). To a certain extent it makes perfect sense to use different PCR primer combinations for amplification of the same gene. In this way one might circumvent the pitfall of not being able to amplify incidental variants of the mef gene carrying a point mutation in the primer-specific region. Unfortunately, not all primer combinations appear to be equally suitable for amplification of the mef genes. In no less than 9 of these 14 PCR primer combinations, the sequences of either or both the forward and the reverse primer do not match both target sequences equally well due to the presence of a mismatch between the primer sequence and one of the two target sequences. If this mismatch is located near the 5′ end of the primer or several residues away from the 3′ end of the primer, this will probably not affect amplification of either or both mef genes. However, in several cases the mismatch(es) is located at the ultimate 3′ end of the primer sequence or multiple mismatches are present along the sequence of the primer(s). In theory, the use of such primers may result in an inefficient PCR, leading to the preferential amplification of the mef gene without the mismatch. In this context, it is entirely conceivable that in the studies in which the primer combination of Oster et al. (62) has been used (19, 62), only the mef(A) gene was reported (as confirmed by restriction enzyme analysis; see below). Likewise, by use of the primer combination of Ono et al. (60), only the mef(E) gene was reported (although this was not confirmed). Interestingly, the primer combination reported by Amezaga et al. (5) in combination with regular Taq DNA polymerase would, in theory, be specific for the mef(E) gene but reportedly amplified both mef(A) and mef(E) (as confirmed by DNA sequence analysis). It is noteworthy that in many studies PCR-based assays were used with selected isolates following a prescreening by use of the erythromycin or the clindamycin MIC. The use of a single PCR primer combination or the use of suboptimal primer pairs may explain why sometimes no explanation for resistance to macrolides could be found (although we cannot ignore the possibility that this could also have been the result of the presence of other known determinants [but which were not tested for] or even yet unknown macrolide resistance determinants). Fortunately, however, the large majority of studies used PCR primer combinations that were able to amplify both mef(A) and mef(E) (Table 3).

TABLE 3.

Overview of PCR primer combinations used for amplification of the mef gene

First author	Forward primer sequencea	Pos.b	Spec.c	Reverse primer sequencea	Pos.	Spec.	Amplicon size (bp)	Assay specificity	No. of reports	Reference
Sutcliffe	AGTATCATTAATCACTAGTGC	57	A = E	TTCTTCTGGTACTAAAAGTGG	402	A = E	346	A + E	50	80
Clancy	CTATGACAGCCTCAATGCG	−313	A = E	ACCGATTCTATCAGCAAAG	1122	A > E	1,435	A + E	8	16
Tait-Kamradt	aaaactgcagGCGTTTAAGATAAGCTGGCd	−297	A = E	ccaatgcatCCTGCACCATTTGCTCCTACd	1462	A > E	1,759	A + E	4	84
Arpin	ATGGAAAAATACAACAATTGGAAAC	1	A = E	TTATTTTAAATCTAATTTTCTAACCTC	1218	E > A	1,218	A + E	3	10
Luna	GGACCTGCCATTGGTGTG	181	E > A	ACCGATTCTATCAGCAAAG	1122	A > E	942	A + E	2	45
Oster	GACCAAAAGCCACATTGTGGA	−102	A	CCTCCTGTCTATAATCGCATG	1329	A = E	1,431	A	2	62
Nagai	CTGTATGGAGCTACCTGTCTGG	288	E > A	CCCAGCTTAGGTATACGTAC	581	E > A	294	A + E	2	57
Marchese	AGTATCATTAATCACTAGTGC	57	A = E	CGTAATAGATGCAATCACAGC	552	A = E	496	A + E	2	48
Shortridge	ATGCAGACCAAAAGCCACCATe	−107	A = E	GCCATAGACAAGACCATCGC	146	A = E	253	A + E	1	78
Ono	ATGGAAAAATACAACAATTGGAAACGA	1	E	TTATTTTAAATCTAATTTTCTAACCTC	1218	E > A	1,218	E	1	60
Farrell	TATGGGCAGGGCAAGCAGTATC	41	A = E	TCRGCACCAATCATTATCTTCTTC	242	A = E	202	A + E	1	27
Amezaga	ATGGAAAAATACAACAATTG	1	A = E	TTATTTTAAATCTAATTTTCTAAC	1218	E	1,218	E	1	5
Marchandin	CTATGCGATTTTGGGACCTG	168	E	GAAAGCCCCATTATTGCACA	968	A = E	801	E	1	47
Woo	ATGGAAAAATACAACAATTGG	1	A = E	TTATTTTAAATCTAATTTTCT	1218	A = E	1,218	A + E	1	92

^aThe sequence of all primers is written in the 5′ to 3′ direction.

^bPos., position. The numbers indicate the positions of the 5′ ends of the primer relative to the position of the ATG start codon (where A is position +1).

^cSpec., specificity. A = E, the primer matches both the mef(A) and the mef(E) gene sequences equally well; A > E, a theoretical preference of the primer for the mef(A) gene; E > A, a theoretical preference of the primer for the mef(E) gene; A or E alone, the primer is specific for mef(A) or mef(E), respectively. The underlined residue(s) in boldface indicates the nucleotide(s) responsible for this preference.

^dThe lowercase letters in the primer sequences were tags added to facilitate cloning of the PCR fragments. Nearly identical primers but without these tags were later used by Del Grosso et al. (23).

^eThis published forward primer contains a typographical error (the underlined C residue printed in boldface should be deleted).

TOOLS TO DISCRIMINATE BETWEEN mef(A) AND mef(E)

After having established the presence of mef genes in clinical isolates, the ability to discriminate between mef(A) and mef(E) naturally depends on the techniques and assays used. The high degree of similarity between the two genes does not allow a reliable discrimination to be made by using DNA hybridization experiments. Unless the probe is meticulously designed, any probe based on the mef(A) gene sequence will also hybridize to the mef(E) gene, even under high-stringency conditions, and vice versa. In contrast, DNA sequence analysis may yield the ultimate means of discrimination between mef(A) and mef(E) (and may even identify point mutations and novel mef-related genes), provided that the proper reference sequences are being used for comparison. In this respect, it should be needless to mention that the accuracies of the DNA sequences deposited in the public libraries are of crucial importance. Here, however, a potential problem arises in the case of mef(A) (see above). The most straightforward method for discrimination between mef(A) and mef(E) is based on the differential presence of restriction enzyme recognition sites in the two genes. A simple digest with a number of restriction enzymes followed by agarose gel electrophoresis should be sufficient to establish the difference. For all of the 14 PCR primer combinations described in Table 3, multiple restriction enzyme recognition sites exist to discriminate between mef(A) and mef(E), which allows the easy dissemination of such an approach. Unfortunately, only only a few groups have applied this approach. A minor caveat to the use of this method is the incidental occurrence of point mutations in the mef gene that could destroy existing restriction enzyme recognition sites or create additional restriction enzyme recognition sites. The occurrence of point mutations in both the mef(A) and the mef(E) genes have been mentioned in several publications (10, 66, 84). The use of more than one restriction enzyme for the identification of each gene may circumvent this pitfall and might even allow recognition of such mutations. As far as is known, this has not led to erroneous conclusions. Furthermore, a real-time PCR assay for the identification of mef genes and discrimination between mef(A) and mef(E) has recently been described by Klomberg et al. (41). This assay enables the even easier discrimination between the two. In conclusion, simple techniques can establish the difference between mef(A) and mef(E).

REVIEW OF LITERATURE DATA: GENES REPORTED VERSUS GENES ANALYZED

In light of the information presented above, the existing literature was reviewed with respect to the subject of the study according to the authors and what the actual subject of the study was according to the methods used: mef(A) or mef(E), or both (the studies are summarized in Table 4). Unfortunately, discrimination between mef(A) and mef(E) was properly established (either by restriction enzyme digestion or by DNA sequence analysis of the PCR amplicons obtained) in only 14 of 77 (18%) publications. In 53 publications, only a PCR was performed to detect the mef gene. In an additional six cases, a PCR was combined with some form of DNA hybridization assay (like a dot blot, Southern blot, or microwell hybridization assay). In four more cases, only a DNA hybridization assay was performed. Consequently, as highlighted in Table 4, in a total of 63 of 77 (82%) publications reviewed, mef(A) and mef(E) could not be discriminated on the basis of the methods used. This is properly acknowledged in 13 cases, in which the target of the assay is generally referred to as either mef or mef(A/E). In the remaining 50 publications, the subject of study is claimed to be either the mef(A) gene or the mef(E) gene, with no effort undertaken to discriminate between the two genes. As mentioned above, in a small number of cases (5, 7, 46, 47, 56, 66, 82) this was done in good faith with reference to the aforementioned recommendation by Roberts et al. (74). In the majority of other studies that reportedly dealt with mef(A), there was no specific reference to the paper by Roberts et al. (74), nor was it apparent from the context that the authors were aware of the nomenclature recommendation. Furthermore, in a number of cases it was assumed that when S. pneumoniae isolates were studied, it must have involved the mef(E) gene (since this gene was first described in S. pneumoniae). Correspondingly, when S. pyogenes isolates were studied, the gene involved was assumed to have been the mef(A) gene. However, we now know that both the mef(A) and the mef(E) genes are present in S. pneumoniae as well as in S. pyogenes and that these working assumptions were actually incorrect. Thus, in the large majority of the publications in which mef(A) or mef(E) only is mentioned, the mef gene detected might just as well have been the other mef gene or even both genes. Interestingly, in six publications, a biphasic MIC distribution was obtained for isolates carrying only a mef gene and no other gene or mechanism that would result in macrolide resistance (21, 32, 37, 54, 59, 87). This indicates that the collection of strains under investigation may have contained two distinct variants of macrolide efflux genes. This observation nicely fits the observed different antibiotic resistance levels for isolates carrying mef(A) and mef(E) (5). In those studies, the populations of strains investigated may very well have contained both mef(A) and mef(E). The isolates for which the MICs were lower may have been carrying mef(E), whereas the isolates for which the MICs were higher may have been carrying mef(A). As a consequence of this, in 82% of the articles in the literature examined, it is confusing, to say the least, whether the actual subject of the publication was mef(A) or mef(E), or both. However, Table 3 provides a convenient means to review these aspects of the available literature.

TABLE 4.

Overview of literature data on detection of mef genes and the ability to discriminate between mef(A) and mef(E)^a

Method(s)b	Subject of publication	Discrimination possible	Reference(s)
PCR	mef	No	3, 21, 22, 24, 28, 40, 44, 70-72, 78, 90
PCR	mef(A)	No	1, 2, 7, 11, 14, 20, 25, 26, 30-34, 36, 39, 50, 51, 56, 57, 61, 64, 65, 67-69, 73, 76, 82, 83, 86, 89, 93
PCR	mef(E)	No	37, 42, 43, 52, 53, 59, 63, 87, 91
Probe	mef(A)	No	77
Probe	mef(E)	No	17, 48, 58
PCR + probe	mef	No	45
PCR + probe	mef(A)	No	27, 38, 46
PCR + probe	mef(E)	No	6, 49
PCR + RE	mef(A) and/or mef(E)	Yes	8-10, 12, 13, 19, 23, 54, 55
PCR + seq	mef(A) and/or mef(E)	Yes	5, 15, 35
PCR + probe + RE	mef(A) and/or mef(E)	Yes	62
PCR + probe + seq	mef(A) and/or mef(E)	Yes	66

^aIn a large number of publications the subject of study is claimed to be either mef(A) or mef(E) when, in fact, on the basis of the methods used, no discrimination between the two could be made.

^bPCR, PCR-based assay; Probe, some form of DNA hybridization assay; RE, restriction enzyme analysis; seq, DNA sequence analysis. In some publications a selected number of strains were further characterized by additional techniques. This is not taken into consideration in this table.

Although Amezaga et al. (5) yielded statistically significant data about the differences in the MICs for strains carrying mef(A) and mef(E); and even though this finding was later confirmed by Neeleman et al. (C. Neeleman, C. H. W. Klaassen, H. A. de Valk, and J. W. Mouton, Abstr. 43rd Intersci. Conf. Antimicrob. Agents Chemother., abstr. C2-68, 2003), it has also been reported that almost all S. pneumoniae isolates confirmed to be carrying mef(A) appear to be highly clonally related (5, 23). This places the MIC differences in an uncertain light, so more work clearly needs to be done to address this issue. Ideally, this could be investigated by inserting the mef(A) and mef(E) genes in an identical defined genetic background. Alternatively, more clonally unrelated strains carrying mef(A) should be tested.

CONCLUSIONS AND RECOMMENDATIONS

Ongoing insight into the properties of the mef genes now acts in favor for maintenance of the difference between mef(A) and mef(E). This does not mean that in certain studies it could be of only minor relevance to determine if a macrolide resistance gene actually represents mef(A) or mef(E). Much of the current confusion about the mef gene nomenclature can be resolved by making the distinction after all. If it had not been for those who took the effort to distinguish between mef(A) and mef(E), we would not have known about the marked differences between them. Furthermore, as outlined above, simple procedures can be used to determine whether one is dealing with mef(A) or mef(E), so it should be easy to establish the difference. In this respect, we must also reflect on the 1999 nomenclature recommendation by Roberts et al. (74). If we were to continue using the nomenclature as suggested, we would be dealing with mef(A) subclass mef(A) and mef(A) subclass mef(E). Not only does this look rather awkward, but more groups already tend to use the original names, mef(A) and mef(E), or use the common name mef instead of following the previously recommended nomenclature suggestion. Although at the time it made sense to suggest a common name for both genes, in retrospect it may be concluded that it has been rather misfortunate to suggest the name mef(A) for both genes instead of the more common name mef. By using the common name mef(A) for all mef genes, much of the current confusion will continue. However, by using the name mef, it is immediately apparent to the reader that no efforts were made to discriminate between mef(A), mef(E) as well as future other variants. Therefore, it makes sense to refer to these genes as just mef in a general context or as mef(A) or mef(E) and other new variants only when appropriate assays have been performed to establish the difference.