The Xmn I site (-158, C----T) 5' to the G gamma gene: correlation with the Senegalese haplotype and G gamma globin expression.

There are three major African haplotypes associated with the sickle mutation: Benin (#19), Senegalese (#3), and Central African Republic (#20). Previous studies have suggested that the Xmn I site (-158 bp 5' to the G gamma gene) is associated with elevated levels of G gamma and with the Senegalese haplotype, while other investigators questioned this association. In order to clarify the issue, we have determined beta haplotypes, tested for the presence of the Xmn I site, and measured Hb F and G gamma expression levels in 143 American Black patients with sickle cell anemia. Haplotypes were determined using eight polymorphic sites in the beta-like globin gene cluster: Hinc II 5' to epsilon, Hind III in IVS-II G gamma and A gamma, Hinc II within and 3' to psi beta, Ava II in IVS-II of beta, and Hpa I and Bam HI 3' to beta. The G gamma /A gamma ratio was analyzed by high performance liquid chromatography using a C18 column. The Xmn I site was present in all 31 chromosomes with the Sengalese haplotype. Of the remaining 255 chromosomes with other haplotypes, only 2 (0.8%) had the Xmn I site present. There was significant correlation between the presence of the Xmn I site and increased G gamma /A gamma ratio in a dose-dependent manner. The Hb F level was not significantly increased in the presence of the Xmn I site. The data indicate that the Xmn I site maintains a G gamma /A gamma ratio typical of fetal life but does not necessarily cause elevation of Hb F. The latter seems to depend on factors other than the Xmn I site.     

Concordance of a point mutation 5' to the G gamma globin gene with G gamma beta +. Hereditary persistence of fetal hemoglobin in the black population

Hereditary persistence of fetal hemoglobin (HPFH) is a genetically heterogeneous and clinically benign condition characterized by persistent expression of fetal hemoglobin (Hb F) into adulthood. In the G gamma beta + type, no major deletions in the globin gene cluster occur; adult heterozygotes produce approximately 20% Hb F, which results from overproduction of G gamma chains, with no apparent increase in production from the adjacent A gamma gene. We have recently described a point mutation 202 base pairs 5' to the cap site of the G gamma gene in an individual with G gamma beta + HPFH. This mutation abolishes a normal ApaI restriction endonuclease site, and thus can be detected by blotting of genomic DNA. We present here further data on the ApaI mutation: (1) It occurs in six of seven families with G gamma beta + HPFH. (2) In three families, detailed haplotype analysis using 11 polymorphic restriction sites in the beta globin cluster has been done. The two that carry the missing ApaI site are identical but the third, which has a normal ApaI pattern, differs from the other two in at least two sites, one of which is a new polymorphic Nco I site between the delta and beta globin genes. This suggests the possibility of a different HPFH mutation in the third family. (3) The haplotype of the G gamma beta + HPFH chromosome carrying the ApaI mutation is different from that of 108 beta A chromosomes of black individuals that have been tested. (4) The G gamma ApaI site is normal in 61 beta A and 109 beta S alleles from non-HPFH black individuals, including 22 who share the same haplotype for the intragenic G gamma, A gamma HindIII polymorphisms. These data add support to the possibility that the -202 mutation is actually causative of the G gamma beta + HPFH phenotype.     

Association of the level of G gamma chain in the fetal hemoglobin of normal adults with specific haplotypes

The levels of G gamma chain in the fetal hemoglobin of more than 40 Black and Caucasian females were determined with a sensitive high performance liquid chromatography procedure and were correlated with their haplotypes, defined by the presence or absence of 10 different restriction sites. Blood was collected during the 16th and 31st week of pregnancy because of a slightly elevated level of Hb F which facilitated the isolation of this protein from a relatively small sample. Four distinct G gamma levels were observed, each being associated with a specific haplotype. Homozygosity for sub-haplotype A [- + + - + +] is associated with high G gamma values (60-70%); that for sub-haplotype B [- - - - - +] with low levels (25-30%); and that for sub-haplotype C [+ - - - - -] with very low levels (10-15%) (restriction sites listed are Hinc II at epsilon; Xmn I 5' to G gamma; Hind III at G gamma and A gamma; Hinc II at psi beta and 3' to it). Sub-haplotype D [(14)- + - - +] with a rare polymorphism 5' to epsilon is associated with extremely high G gamma values. Hb F levels were low (less than 2.5%) and were independent of the haplotype. It is speculated that, yet unknown, variations in the DNA of gene activity controlling regions are responsible for the differences in G gamma value.     

The -158 site 5' to the G gamma gene and G gamma expression

To test the hypothesis advanced by Gilman and Huisman that the -158 site 5' to the G gamma gene determines the G gamma expression after the first 4 months of life, we have examined DNA from sickle cell anemia (SS) patients from Africa and beta-thalassemic homozygotes from Algeria. We find that the Xmnl site is strongly linked to the Senegal haplotype among SS patients, to haplotype IX (most probably identical to the Senegal haplotype), and to haplotype III among the Algerian thalassemics. Thalassemics with haplotypes I/I and V/V have no Xmnl site and low G gamma expression. In contrast, beta-thalassemia- associated haplotype II (also characterized by high G gamma expression) fails to exhibit the Xmnl site. We conclude that, although highly correlated, the -158 C----T substitution does not perfectly predict the presence of high G gamma expression. These findings also exclude the possibility that the Xmnl site is solely involved in the determination of high G gamma expression and suggest that either several different site substitutions in the area 5' to the gamma gene might have the same effect or that, alternatively, the Xmnl site and its surrounding area is not involved in G gamma expression and may be only in linkage disequilibrium with a controlling sequence elsewhere.     

Beta-cluster haplotypes, alpha-gene status, and hematological data from SS, SC, and S-beta-thalassemia patients in southern California

The beta-gene-cluster haplotype and alpha-gene status were determined for 221 patients with sickle cell anemia, 41 with SC disease, and 21 with S-beta-thalassemia. Among SS patients, eleven beta S haplotypes were found in 21 combinations. Three haplotypes--the Benin (Ben) [---+-], the Central African Republic (CAR) [+---+], and the Senegal (Sen) [+- ]--comprise 61%, 21%, and 10% of the chromosomes, respectively. Cleavage at the Xmn I site 5' to the G gamma gene was observed only when the Senegalese arrangement was present. The linear correlation which exists between the absolute value of the G gamma chains and the Hb F for each haplotype combination suggests a feed-back mechanism which controls the G gamma to A gamma ratio and thus the Hb F level (or vice versa). The A gamma T chain was present with specific haplotypes [++-++] and [++-+-]. Heterozygous or homozygous alpha-thalassemia-2 was present in 36% of the SS patients and was randomly distributed among beta S-gene-cluster haplotypes. The variable levels of hemoglobin, MCV, Hb F, G gamma chains, and Hb A2 are in response to the heterogeneous genetic mix of the beta S-gene-cluster haplotypes and alpha-thalassemia-2 in American patients with sickle cell anemia. The influence of alpha-thalassemia-2 on the level of Hb F is dependent on the beta S-cluster haplotype. Hb A2 levels increased with decrease in the number of alpha genes. Among SC and S-beta-thalassemia patients the beta-cluster polymorphisms on the beta S chromosome were those commonly associated with the African origins of beta S haplotype. The haplotype [+--+-] was present on the C chromosome in 90% of the cases. Most beta-thalassemia chromosomes had haplotypes that matched the common African polymorphisms. An alpha-gene deletion was found in 29% of the SC and S-beta-thalassemia patients.     

DNA sequence variation associated with elevated fetal G gamma globin production

The percentage of G gamma chains in the Hb F of SS patients and beta-thalassemia heterozygotes is generally 40%, but some have 60% to 70% G gamma. To test the hypothesis that DNA sequence variation 158 base pairs 5' of the G gamma gene is associated with this variation in G gamma values, DNA was analyzed using the restriction endonuclease Xmn I (gamma IVS-II probe). Xmn I recognizes the sequence from -157 to -166 only if T is at position -158. Individuals from five families had T at -158 for G gamma genes in both chromosomes, and the mean G gamma value was 69.7% +/- 4.6% (SD). For 13 families, individuals with T at -158 for the G gamma gene of one chromosome had a G gamma value of 60.6% +/- 5.7%. With one exception, lack of T at -158 was associated with low G gamma values (39.6% +/- 4.0%). In low Hb F G gamma-beta+-HPFH, the Xmn I site was seen 5' to both G gamma and A gamma genes, which suggests that T at -158 is associated with elevated Hb F; Pst I digestion showed that the A gamma gene T producer G gamma globin, which accounts for high levels of G gamma (87-88%). Calculations show that T at -158 is associated with a three- to 11-fold increase in production per G gamma gene, which is an order of magnitude less than that associated with the previously identified -202 C----G substitution of high Hb F G gamma-beta+-HPFH.     

Four categories of gamma-globin gene triplications: DNA sequence comparison of low G gamma and high G gamma triplications

The human fetal gamma chains are produced by closely linked G gamma and A gamma genes, and unequal crossing over between them leads to gamma gene deletions and triplications. Nine gamma gene triplications from seven ethnic groups were analyzed for G gamma and hemoglobin F (Hb F) values of heterozygotes and for the presence of polymorphic XmnI restriction sites 5' to the gamma genes. Four categories of triplication were found: I had low G gamma and low Hb F values and lacked XmnI sites 5' to the three gamma genes [---]. II had high G gamma and slightly elevated Hb F values but was also [---]. III was similar to II, except that XmnI was [+--]. IV had very high G gamma and slightly elevated Hb F values, and XmnI was [++-]. One case each of triplications I and IV were cloned into Charon 35. For both, the two 5' gamma gene code for G gamma chain, while the 3' gamma gene codes for A gamma chain. DNA sequencing showed that the unequal crossover occurred between 472 and 398 base pairs (bp) 5' to the gamma gene Cap sites (- 472 and -398) for the type IV triplication and between -271 and codon 136 for the type I triplication. In addition, type I had a 4-bp deletion of AGCA from -225 to -222. The high G gamma values of the type IV triplication are explained by its -G gamma-G gamma-A gamma-gene arrangement and the XmnI sites 5' to the G gamma genes. We hypothesize that the low G gamma value of the type I triplication, which is also -G gamma-G gamma-A gamma-, is due to inactivation of the middle G gamma gene by the AGCA deletion at -225 to -222.     

Haplotypes of beta S chromosomes among patients with sickle cell anemia from Georgia

Fetal hemoglobin and G gamma levels have been correlated with the presence or absence of eight restriction sites within the beta globin gene cluster (haplotypes) for numerous sickle cell anemia patients from Georgia. The most common haplotypes were #19 (Benin) and #20 (CAR); all patients with haplotype combinations 19/19, 20/20, and 19/20 were severely affected with low Hb F and low G gamma levels. A modified #19 beta S chromosome with a -G gamma-G gamma- globin gene arrangement, instead of -G gamma-A gamma-, was present in SS and SC newborn babies with G gamma values above 80%. Haplotype #3 (Senegal) was present among 15% of the beta S chromosomes; the two adult patients with the 3/3 combination were mildly affected with high Hb F and G gamma values. The haplotype AT with the variant A gamma T chain was a rarity. A new haplotype was found in one 17-year-old SS patient and five of his Hb S heterozygous relatives. This haplotype is associated with an increased production of Hb F in heterozygous and homozygous Hb S individuals; this Hb F contained primarily A gamma chains. A comparison was made between the different haplotypes among SS patients and normal Black individuals, and a remarkable similarity was noted in the fetal hemoglobin data for subjects with these different chromosomes.     

Hematological observations on Arabian SS patients with a homozygosity or heterozygosity for a beta S chromosome with haplotype #31

Hematological and hemoglobin composition data are presented for seven Arabian SS patients with mild disease and with high Hb F levels varying between 21 and 34%. Four patients were homozygous for a beta S chromosome with a specific haplotype (#31). The data for these four patients were similar to those for three other SS patients (and for five patients reported earlier, Ref. 2) who were heterozygous for the same beta S chromosome (#31) and for a beta S chromosome with another haplotype (mainly #19). These data offer additional evidence indicating that the increased gamma chain production is specific for the beta S chromosome with haplotype #31. The similarities in hematological data and Hb F levels between these two groups of SS patients and the normal Hb F value in Hb S heterozygotes with beta S chromosome (#31) support the suggestion that the increased Hb F production mainly occurs in response to the anemia of the sickle cell disease.     

Effect of beta-globin gene cluster haplotype on the hematological and clinical features of sickle cell anemia

In 113 black American adults with sickle cell anemia (HbSS), we examined nine polymorphic restriction sites, including the Xmnl site 5' to the G gamma gene, to see whether haplotype is related to the level of HbF and the proportion of G gamma chains or if it influences the hematological and clinical features of the disease. Seventy-five percent of the patients were homozygous or heterozygous for the Benin (no. 19) or Central African Republic (Bantu, no. 20) haplotypes; 13.3% were homozygous or heterozygous for the Senegal (no. 3) haplotype, while 11.5% had other genotypes. Of the subjects, 14.2% were either homozygous or heterozygous for the Xmnl restriction site 5' to the G gamma gene. We found no effect of haplotype on HbF levels. The level of G gamma chains was 60.5% +/- 17.0% in individuals heterozygous or homozygous for haplotype no. 3 and was 46.9% +/- 11.6% in individuals with other haplotypes. Subjects with the Xmnl site 5' to the G gamma gene had G gamma globin levels of 59.5% +/- 16.7% while those lacking that site had an average of 47.2% +/- 12.1%. There were no significant differences among these groups in hemoglobin concentration, packed cell volume, mean cell volume, or clinical indicators of vaso-occlusive severity, including crises, hospitalizations per year, aseptic bone necrosis, acute chest syndrome, or leg ulcers. While the presence of haplotype 3 and the 5' G gamma Xmnl site were associated with increased G gamma chains, there was no effect on HbF level or other hematological and clinical features that might reflect disease severity. It is likely that determinants unrelated to haplotype, linked or unlinked to the beta-globin gene cluster, are the major effectors of differences in the levels of HbF in American patients with sickle cell anemia.     

Homozygosity for a new type of G gamma (A gamma delta beta)zero-thalassemia in a Malaysian male

Hematological and clinical data are presented for a young Malay patient with a homozygous (delta beta)zero-thalassemic condition. His red blood cells contained 100% fetal hemoglobin with alpha and G gamma chains only. Detailed gene mapping defined a large deletion with a 5' end between the Aha III and Apa I sites, some 200-400 bp 5' to the A gamma globin gene and a 3' end beyond sequences 17-18 kb 3' to the beta globin gene. This G gamma (A gamma delta beta)zero-type of thalassemia is different from all the other six types described before. Comparison of the hematological data of this patient with those of homozygotes for either the Sicilian or Spanish types of G gamma A gamma (delta beta)zero-thalassemia showed no differences; all homozygotes have a moderate anemia which is accentuated by the relatively high oxygen affinity of the Hb F containing erythrocytes.     

DNA polymorphism in the beta-globin gene cluster in Saudi Arabs: relation to severity of sickle cell anaemia.

Significant DNA polymorphisms have been reported in the beta-globin gene cluster of epsilon-G gamma-A gamma-psi beta-delta-beta-gene region, in normal (Hb AA) individuals and in patients with sickle cell anaemia (SCA). Investigations of the extent of the DNA polymorphisms in the beta A- and beta S-globin gene cluster using Hind III, Hinc II, Ava II, Xmn I, and Hpa I, revealed several associations with mild SCA. The correlation of the presence (+) or absence (-) of the restriction endonuclease site to clinical severity in patients homozygous for beta S-gene showed that the mild form of SCA was associated mainly (> 90%) with the Xmn I polymorphic site 5' to G gamma, and to a lesser extent with Hinc II polymorphic site 5' to epsilon and in the psi beta-gene, with Hind III polymorphic site in G gamma and Hpa I polymorphic site 3' to the beta-globin gene, while in the severe form of SCA these polymorphic sites were absent in most patients. The polymorphism in the beta-globin gene cluster was significantly related to the expression of the beta S-gene and clinical severity of SCA.     

Haplotypes and levels of fetal hemoglobin and G gamma to A gamma ratios in Mediterranean patients with thalassemia minor and major

This study concerned the gamma chain composition of Hb F and the haplotypes of 44 patients with beta-thalassemia major or intermedia and many of their relatives. Seventeen patients came from Northern (Turkish) Cyprus, 12 from the Istanbul area, and 15 from Macedonia and Bulgaria. Analysis of the A gamma T-G gamma-A gamma I ratio was made by HPLC, while haplotyping involved seven restriction sites. Specific haplotypes were present in certain populations; haplotype I [1] is the dominant type among North Cypriot thalassemia patients. Numerous types were seen in the patients from the Balkan countries. A direct relationship between the A gamma to G gamma ratios and the haplotypes, which exists among black beta-thalassemia heterozygotes [3], was also observed among these Mediterranean patients, although such analyses were considerably complicated by extensive blood transfusion therapy. Haplotypes without the Hinc II restriction site within the psi beta gene were associated with lower G gamma values than those that had this polymorphic site. The A gamma T chain was observed in a small number of beta-thalassemia homozygotes and heterozygotes. Three thalassemia chromosomes with slightly different haplotypes and one normal chromosome with a related haplotype were associated with the gamma 75 Ile----Thr substitution. A few patients with a thalassemia intermedia were heterozygotes for beta-thalassemia with either haplotypes V or VII [1] while the "nonthalassemic" chromosome had a haplotype I, which is the most common "beta-thalassemic" haplotype among the Mediterranean population(s). Detailed analyses of this chromosome have not been completed.     

Hemoglobin abnormalities in a black family with HB S, hereditary persistence of HB F, and a gamma chain variant; a reevaluation through gene mapping

Members of a Black family from Georgia who were investigated for the first time in 1960 and several times thereafter were reinvestigated through DNA restriction endonuclease analyses and haplotyping, while the gamma chain heterogeneity of the Hb F was reevaluated using a newly developed HPLC procedure. Four different abnormalities were present. (a) Heterozygosity for G gamma A gamma-HPFH type II characterized by a large deletion involving the delta and beta globin genes with a 5' end within the psi beta gene. (b) Heterozygosity for an -epsilon-G gamma-G gamma-psi beta-delta-beta S-chromosome, thus carrying a beta S globin gene and two G gamma genes instead of one G gamma and one A gamma gene. (c) Heterozygosity for an -epsilon-G gamma-A gamma T-psi beta-delta-beta S-chromosome, carrying the beta S globin gene and an allele of the A gamma (or A gamma I) gene. These three chromosomes occurred in combination with each other, resulting in SS and S-HPFH conditions, and with a normal -epsilon-G gamma-A gamma-psi beta-delta-beta A-chromosome resulting in the HPFH and Hb S heterozygosities. The presence of the -G gamma-G gamma- and -G gamma-A gamma T-chromosomes in the one SS patient was responsible for the high G gamma value (average 75%), 25% A gamma T chain, and for the absence of the A gamma I chain. (d) An alpha-thalassemia-2 heterozygosity in one member.     

An A gamma type of nondeletional hereditary persistence of fetal hemoglobin with a T----C mutation at position -175 to the cap site of the A gamma globin gene

The nondeletional types of hereditary persistence of fetal hemoglobin (ndHPFH) concern the continued synthesis of hemoglobin (Hb) F with either G gamma or A gamma chains in amounts varying from 5% to 30%. Several mutations have been identified in either the A gamma or G gamma promoter which are considered causative to the continued production of one of the two gamma chains because the substitutions occur in sequence motifs essential for the expression characteristics of the gamma-globin gene in the 3' position. We report the discovery of a T----C mutation at position -175 in the A gamma promoter which was associated with a greatly increased level of Hb F (with mainly A gamma) and a decreased level of Hb A in the one (Black) heterozygote who had a beta c gene in trans. The same mutation has been observed in the G gamma promoter of a Black heterozygote who had high levels of Hb F with G gamma chains only. A detailed comparison between these two individuals indicated significant differences in the levels of Hb F and Hb A which may result from an additional mutation at position -158 in the G gamma promoter.     

Certain mutations observed in the 5' sequences of the G gamma- and A gamma-globin genes of beta S chromosomes are specific for chromosomes with major haplotypes.

In this paper we describe the distribution of some specific sequence differences in the 5' flanking regions of the A gamma- and G gamma-globin genes from 100 Black adult and 57 newborn SS patients from the southeastern United States, from 76 individuals with AS, S-beta-thal, SC, AC, or A-beta-thal, and from 31 normal individuals. Haplotypes for all adult individuals have been previously determined using various restriction endonucleases. The DNA samples were amplified, dot blotted, and hybridized with 32P-labeled specific oligonucleotide probes. All 134 chromosomes with haplotype 19 were positive for the G----T substitution at position -657 (A gamma), while 132 were also positive for the C----G mutation at -369 (G gamma). The three specific changes for the chromosome with haplotype 20 were found on all 54 chromosomes with this haplotype. The C----T mutation at -158 5' to G gamma was present on all 41 chromosomes with haplotype 3, and on two chromosomes with a related atypical haplotype. Normal and beta-thal chromosomes with each of these substitutions had the same 5' subhaplotype as beta S haplotypes 19 or 20, respectively. The close relationship between the occurrence of specific mutations and the haplotype of beta S chromosomes makes the determination of these haplotypes with specific oligonucleotide probes attractive with respect to time and expense.     

Association of thalassaemia intermedia with a beta-globin gene haplotype

We have identified 14 Asian patients with homozygous beta zero thalassaemia who had a mild clinical disorder related to an augmented production of haemoglobin F. None of their parents had an elevated level of Hb F. Restriction fragment length polymorphism analysis of the beta-globin cluster of these patients and a control group of Asian thalassaemia major patients showed that 6/14 of the thalassaemia intermedia patients were homozygous for a particular 5' beta-globin haplotype (-+-++), in contrast to 1/42 of the thalassaemia major patients. Furthermore, the -+-++ beta haplotype is also associated with amelioration of disease severity in beta thalassaemia in an Italian population. This beta haplotype is linked to a DNA sequence variation 5' (at position -158) to the G gamma globin gene which can be detected by the presence (+) of an Xmn I restriction enzyme site. The possible role of the Xmn I-gamma polymorphism in relation to this variant HPFH is discussed. We conclude that much of the observed clinical variability of beta thalassaemia can now be explained by the inheritance of beta thalassaemia chromosomes with different propensities for fetal haemoglobin production.     

Two independent genetic factors in the beta-globin gene cluster are associated with high G gamma-levels in the HbF of SS patients

The gamma-chains of fetal hemoglobin (HbF) of newborn babies are composed of about 70% G gamma and 30% A gamma. In most babies, the G gamma value declines postnatally to 40%, but in about 20% of black SS patients from Georgia, 5 years and older, the G gamma level remains high at 60%. Moreover, some 3% to 4% of black newborns have high G gamma values of 85%. PstI digestion of DNA of one such high G gamma baby and of one normal newborn showed the former to be heterozygous for the -G gamma-G gamma- and -G gamma-A gamma-chromosomes. Only about one fourth of high G gamma SS patients were such heterozygotes, while three fourths were -G gamma-A gamma-/-G gamma-A gamma-homozygotes. Analysis of DNA of 38 SS patients without the -G gamma-G gamma-chromosome showed a correlation of G gamma values with genotype at one polymorphic restriction site: at the HincII site in the psi beta gene, all -G gamma- A gamma-/-G gamma-A gamma-homozygotes with high G gamma were +/- or +/+, while low G gamma individuals were all -/-. Family studies, involving analyses at four polymorphic sites (HindIII sites in the G gamma and A gamma genes and HincII sites in the psi beta gene and 3' to it), suggested the association of an unidentified high G gamma genetic determinant with haplotype + - + +. This indicates that a genetic factor causing high G gamma levels in SS patients is closely linked to the -G gamma-A gamma-psi beta region of the beta-globin gene cluster.     

Expression of G gamma and A gamma globin genes in human adults

Expression of the human fetal G gamma and A gamma globin genes declines shortly after birth, and adults generally have less than 1% fetal hemoglobin or Hb F (alpha 2 gamma 2). However, some adults with hereditary persistence of fetal hemoglobin (HPFH) have elevated expression of either the G gamma or A gamma gene due to a mutation in its upstream promoter. Mutations with strong effects on expression have been found at -175 and -202 of the G gamma gene and at -117, -196, -198 and -202 of the A gamma gene. Mutations at -158 and -161 of G gamma have weaker effects, which are observable primarily as increases in the G gamma:A gamma ratio. Published data are reviewed which suggest that the -158 mutation may lead to observable elevations of Hb F in SS and beta(0)-thal patients and occasionally in normal non-anemic individuals. These data also suggest that additional high Hb F determinants are linked to Benin, Bantu and Asian beta S haplotypes in some instances. A model based on data from SV40 is presented which suggests that specific DNA sequence motifs of the gamma globin gene may bind regulatory proteins. It is proposed that the -158 and -161 mutations have weak effects because they are located on the fringe of regulatory sequence motifs.