Evolution and molecular interactions of major histocompatibility complex (MHC)-G, -E and -F genes

Classical HLA (Human Leukocyte Antigen) is the Major Histocompatibility Complex (MHC) in man. HLA genes and disease association has been studied at least since 1967 and no firm pathogenic mechanisms have been established yet. HLA-G immune modulation gene (and also -E and -F) are starting the same arduous way: statistics and allele association are the trending subjects with the same few results obtained by HLA classical genes, i.e., no pathogenesis may be discovered after many years of a great amount of researchers’ effort. Thus, we believe that it is necessary to follow different research methodologies: (1) to approach this problem, based on how evolution has worked maintaining together a cluster of immune-related genes (the MHC) in a relatively short chromosome area since amniotes to human at least, i.e., immune regulatory genes (MHC-G, -E and -F), adaptive immune classical class I and II genes, non-adaptive immune genes like (C2, C4 and Bf) (2); in addition to using new in vitro models which explain pathogenetics of HLA and disease associations. In fact, this evolution may be quite reliably studied during about 40 million years by analyzing the evolution of MHC-G, -E, -F, and their receptors (KIR—killer-cell immunoglobulin-like receptor, NKG2—natural killer group 2-, or TCR-T-cell receptor—among others) in the primate evolutionary lineage, where orthology of these molecules is apparently established, although cladistic studies show that MHC-G and MHC-B genes are the ancestral class I genes, and that New World apes MHC-G is paralogous and not orthologous to all other apes and man MHC-G genes. In the present review, we outline past and possible future research topics: co-evolution of adaptive MHC classical (class I and II), non-adaptive (i.e., complement) and modulation (i.e., non-classical class I) immune genes may imply that the study of full or part of MHC haplotypes involving several loci/alleles instead of single alleles is important for uncovering HLA and disease pathogenesis. It would mainly apply to starting research on HLA-G extended haplotypes and disease association and not only using single HLA-G genetic markers.

464 Page 2 of 18 was first considered to be an immune modulatory molecule, predominantly expressed at the maternal-fetal interface and its function was first assigned to maternal-fetal tolerance [2,[4][5][6]. Initial studies were carried out by Dan Geraghty et al. [7] and they named HLA-6.0 the new gene they described. HLA-6.0 protein was structurally similar to HLA-A, -B, and -C class I molecules but with a premature in-frame stop codon that hindered translation of an important part of the cytoplasmatic region in HLA-6.0 mature molecule. The promoter region of HLA-6.0 gene was similar to that of MHC-Qa mouse gene, and both genes were equivalent with regard to substitutions, deletions and other variations in allelic DNA sequences [7]. Warner et al. group [8] proposed that MHC-Qa was a functional HLA-G homologue in mouse, with a similar gene and protein structure; MHC-Qa also presents soluble forms like HLA-G5, G6 and G7 isoforms in humans (Fig. 1). Recently, it is found that Qa-1b(MHC-Qa non-classical class I gene in mouse) seems to be homologous to HLA-E (see HLA-E "Evolution" section). The complete HLA-G molecule has an extracellular structure very similar to that of the classical HLA molecules, though its major function is not antigen presentation. It was found that HLA-G inhibits the cytotoxic activity of T CD8 + and NK cells through direct interaction with leukocyte receptors, such as LILRB1 (LIR1/ILT2), LILRB2 (ILT4), and KIR2DL4 (CD158d) [3,[9][10][11][12][13][14].
HLA-G gene and molecule expression patterns differ in many aspects compared to classical HLA class I molecules, like: (a) a restricted tissue expression in normal conditions [21]; it is being expressed on the maternal-fetal interface in the extravillous cytotrophoblast cells [6], cornea, proximal nail matrix, thymus, hematopoietic stem cells and pancreas mainly [22][23][24][25][26][27]. HLA classical class I molecules (HLA-A, -B, and -C) are widely expressed in all body tissues. Nonclassical class I HLA molecules (HLA-E, -F, and -G) are more restricted regarding tissue localization, antigen presentation, and function [3]. Diversity of presented peptides compared with that of classical class I MHC molecules is much reduced probably because of their limited levels of polymorphism [28]. These non-classical class I molecules HLA-G also presents stop codons in intron 2 or intron 4 depending on alternative splicing process which gives rise to different isoforms. Stop codon may be maintained in mature mRNA due to a reading-through mechanism in humans and primates which is described also in other HLA genes (i.e., HLA-DRB6). The presence of a selenocysteine insertion sequence (SECIS) at the 3 untranslated region leads to a selenocysteine incorporation at UGA (stop) codons [15][16][17][18]; this may be the cause for stop codon maintenance in HLA-G, -E and -F translation. Beta-2 microglobulin (β2m) is represented bound to protein molecules in purple color. See also references [19,20] may also regulate immunity through TCR-independent interactions (see below); (b) they show several membrane and soluble isoforms due to alternative splicing of the complete HLA-G mRNA [2,3]; (c) a short cytoplasmic tail is present due to the presence of a premature stop codon at exon 6 [2,3]; (d) a relatively low HLA-G protein polymorphism is recorded although it is rapidly increasing (Fig. 2) [2,3,29]; (e) they present a unique 5'URR (5' upstream regulatory region) different from other HLA classical class I genes [30,31]; and (f) the 5' promoter region [2,[32][33][34][35][36] and the 3'UTR (3' untranslated region) show several polymorphisms that are specifically linked to diseases susceptibility [37].
Also, it has been shown that HLA-G presents endogenous peptides at the surface of the placenta trophoblast [42], Fig. 2 HLA-G protein alleles. Codon and aminoacidic changes among different alleles in exon 2, exon 3 and exon 4 are shown. The letter "N" at the end of some alleles shown in the table denotes null allele. These null alleles bear a stop codon due to single-base deletions or point mutation which give rise to an incomplete HLA-G protein translation. HLA-G*01:05N has a single cytosine deletion at codon 130 (CTG → TGC) which produces a reading frameshift change, causing a premature stop signal at codon 189 (GTG → TGA) [38,39] and consequently a shorter protein with α1 functional domain at least [38,40]. HLA-G*01:21N has a premature stop codon due to a punctual mutation in codon 226 (CAG → TAG) of coding sequence which leads to a non-complete translated protein [40]. The number of HLA-G protein alleles is rapidly growing; see IMGT-HLA database to be up to date on new alleles (https:// www. ebi. ac. uk/ ipd/ imgt/ hla; accessed September 2021) [41] 464 Page 4 of 18 absent in other HLA classical class I molecules' expression [43], with the exception of HLA-C [44]. Thus, HLA-G interacts at this maternal-fetal interface with activating and inhibitory receptors: killer-cell immunoglobulin-like receptor (KIR), leukocyte immunoglobulin-like receptor (LIR), and CD94-NKG2 receptor complex families to establish maternal tolerance and normal fetal growth [43]. This nonclassical class I HLA molecule recognizes TCR of regulatory [45] and cytolytic [46] CD8 T cells [47].
On the other hand, HLA-E polymorphism is represented only by two functional molecules that present a set of similar peptides derived from class I leader sequences. However, HLA-E is a ligand for the innate and adaptive immune system effectors; immunological response to peptide-HLA-E complexes is determined by the sequence of the bound peptide, which interacts with CD94/NKG2 or T-cell receptor [48,49].
While HLA-E and HLA-G have been well-characterized functionally and structurally, the role that HLA-F plays in regulating the immune system has long time been unknown. However, HLA-F has been shown to protect fetus development [50] and has a role at peripheral nervous system: HLA-F recognition by the inhibitory KIR3DL2 receptor prevents motor neuron death in amyotrophic lateral sclerosis physiopathology [51]. Also, HLA-F interacts with the activating KIR3DS1 on NK cells and induces an anti-viral response against HIV-1 (human immunodeficiency virus-1) [52]. HLA-F immunity regulation by KIR3DS1 interaction has increased the clinical importance of HLA-F since also other diseases exist where KIR3DS1 has a pathogenetic role [53]. Thus, HLA-F and disease relationship is important but the molecule structural and biochemical properties and the precise relationship with its function is mostly unknown. "In-silico" studies predicted that HLA-F has the typical MHC fold but with only a partially open-ended groove [47,54].

Role of MHC-G, -E and -F as immune-regulation proteins: pathology
Expression of HLA-G has been studied in autoimmune and inflammatory diseases, tumors, chronic viral infections and in engrafted tissues [5,[55][56][57][58]. This HLA-G expression has been associated with better prognosis in chronic inflammation, autoimmune diseases, and allotransplants, because inhibition of immune response occurs; however, this inhibition may be harmful in chronic viral infections and tumors, where an efficient immune response may be hindered [59,60]. The role and pathology of MHC-G, -E, and -F in maternal/fetal relationship has been widely reviewed [3] (see below), but this must be complemented by HLA-C role, which is the only classical class I molecule expressed at the cytotrophoblast and shows both presenting and suppressive functions [44].

Evolution
Parham et al. studies on classical MHC genes structure and evolution in apes should be consulted to better understand non-classical class I genes evolution [65,66]. New World monkeys lineage separated about 35 million years ago [67,68] from the lineage that gave rise to Old World and anthropoid monkeys. The cotton-top tamarin (Saguinus oedipus, Saoe) that inhabits Central-South America is a typical example of this group and has MHC-G-like genes instead of MHC-A and MHC-B genes [69]. However, MHC-C sequences have been also described in this New World monkey [70], which also binds KIR [71]. MHC of cotton-top tamarin shares more primary DNA sequence homologies with HLA-G than with classical class I HLA genes [69,72,73]. This is why, MHC-G has been assigned as the ancestral MHC class I gene and that MHC class I genes of the Saoe could be homologous to HLA-G genes. MHC-G is also present in Old World Monkeys, although MHC-E primary DNA structure may be closer to that of Saoe MHC [3] (see "HLA-E" section). The α1 domain of MHC-G molecule is preserved in all species studied (Fig. 3) and may be sufficient for MHC-G function in the subfamily of Cercopithecinae monkeys (Macaca mulatta, Macaca fascicularis, Cercopithecus aethiops) [3]. All the MHC-G alleles of this subfamily bear stop codons (like some human individuals; see below in "HLA-G*01:05N, -G*01:01 and -G*01:04 alleles World distribution: significance" section, HLA-G null alleles frequencies distribution) in a very restricted area of exon 3 (at codon 164), and some alleles may also show stop signals at codons 133, 118, and 176 [74]. However, pregnancies are normal in these Cercopithecinae species and functional MHC-G molecules may exist lacking the α2 domain, because one of the most important roles of MHC-G is preserving the fetus from maternal NK cells attack. Otherwise, readingthrough stop codon mechanisms may exist [75]. MHC-G polymorphism is low in the Pongidae family: gorillas and chimpanzees [3,76]. Intron 2 of MHC-G sequences show conserved motifs in all primate species: a 23-bp deletion starting in position 161, which is MHC-G locus specific. Surprisingly, the Saoe MHC-G intron 2 does not bear this deletion. Explanations for this finding could be that: (1) the MHC-G-like sequences in Saoe described did not give rise to the Old World monkey and human MHC-G alleles; or (2) the 23-bp deletion most likely occurred after separation of the New World monkeys from Old World monkey lineages about 35 million years ago [68,69]. The first hypothesis is more plausible, since eluted peptides from cotton-top tamarin MHC-G like molecules are not typical of MHC-G [77]. MHC-G orthology has been studied by simple resemblance phylogenetic comparisons. However, lineal time inferences of species separation may be wrong and interpretation needs caution: this is because of the frequent birth and death processes of genes and/or parts of them observed in the MHC region. Also, MHC-G in New World monkeys turns up as paralogous rather than orthologous to other primate MHC-G genes by cladistic studies on Alu and L1 elements insertions at 5' region [78]. Indeed, this cladistic analysis concluded that MHC-B and MHC-G genes are ancestral to other MHC class I genes.
On the other hand, it has been found that MHC-G4, G5, and G6 isoforms are not present in gorilla, chimpanzee, and orangutan [76]. This finding suggests that MHC-G4 and the G5 and G6 soluble isoforms may be human-specific, and that MHC-G could have evolved independently in each group of primate species. With regard to these new findings, they make more difficult to assign a universal function for primate MHC-G proteins at the placental level or even at controlling autoimmunity [76]. Also, it has been found that MHC-G polymorphism shows more differences in Cercopithecinae family and in Pongidae species: (1) Cercopithecinae family bears a stop codon at exon 3, which is absent in Pongidae family. The latter bears a stop codon in exon 6, like humans [74]. This variation was generated 33 million  [79,80]; (2) exon 7 is not found in MHC-G transcripts in human and Pongidae species, but it is preserved in rhesus monkeys (Cercopithecinae family) MHC-G mature mRNAs [81]; (3) MHC-G2 "short" unusual splice variants have been found in Gorilla (Pongidae) and also in rhesus monkeys (Cercopithecinae) [76]. It seems that during the last 40 million years, a selective pressure has operated on MHC-G protein binding domain (antigen cleft, at exons 2 and 3) in New World and Old World primates and also in humans [15,16].
In summary, it is striking that: (a) HLA-G*01:05N homozygous individuals there exist (non-functional HLA-G1 membrane-bound isoform) [82]; (b) MHC-G4, G5, and G6 isoforms are not necessary for survival in Pongidae family [76]; (c) Cercopithecinae family bears a stop codon at exon 3 [74]. These observations may lead to the conclusion that MHC-G is not a functional protein in Old World monkeys or may be substituted by other molecules [3,64].
Moreover, presence of different HLA-G proteins in different primate species may be evolutionary better explained by mutations (i.e., deletions) that occurred at different apes speciation times. See reference [68], Fig. 3.

DNA transcription and translation
HLA-G exon 1 encodes for the signal peptide. Exons 2, 3, and 4 transcribe for extracellular α1, α2, and α3 domains, respectively; and exons 5 and 6 for the transmembrane and the heavy chain cytoplasmic domain. HLA-G has a short cytoplasmic domain, because there exist a premature stop codon in exon 6; thus, exons 7 and 8 are not transcribed in the mature mRNA [4,5].

Surface molecules
Seven HLA-G transcripts produced by alternative mRNA splicing exist. Four of them give rise to membrane-bound protein isoforms and there are also three soluble isoforms [83]. HLA-G1 isoform is a complete HLA class molecule, with β2-microglobulin association. HLA-G2 lacks the α2 domain encoded for by exon 3 (Fig. 1), and HLA-G3 isoform has neither α2 nor α3 domains, encoded by for exons 3 and 4, respectively (Fig. 1). HLA-G4 does not have α3 domain, encoded by for exon 4. HLA-G5 and HLA-G6 soluble isoforms have the same domains than those of HLA-G1 and HLA-G2 isoforms; they are originated by transcripts which preserve intron 4, hindering the translation of the transmembrane domain (exon 5) (Fig. 1). Intron 4 is translated up to a stop codon in its 5'region; this is the cause that HLA-G5 and HLA-G6 isoforms to have a tail of 21 amino acids accounting for their solubility. HLA-G7 isoform has only the α1 domain together with two amino acids coded by intron 2, which is transcribed [83] (Fig. 1).

Receptors
HLA-G extracellular domains bind to the following leukocyte receptors: CD8, LILRB1 and LILRB2 and the killercell immunoglobulin-like receptor KIR2DL4 (CD158d) (see Table 1). LILRB1 and LILRB2 also interact with the HLA-G molecule α3 domain and β2-microglobulin, LILRB2 having a higher affinity than LILRB1 for the molecule [3]. LILRB-binding sites are different for each receptor [3]. CD8 molecule also interacts with all MHC class I molecules through α3 domain of classical and non-classical MHC-I molecules, like HLA-G and HLA-E. CD8α/α binds to HLA-G with higher affinity, and with a lower affinity to HLA-E [3]. Moreover, β-2 microglobulin binds HLA-G isoform dimers (G1 and G5) and interacts with LILRB1 and LILRB2 receptors; LILRB1 predominantly binds β2-microglobulin-associated isoforms, while LILRB2 preferentially contacts β2-microglobulin-free HLA-G. Ability of HLA-G isoforms to associate in homodimers and their binding affinity depending on the receptor are important for HLA-G function [64,84,85].

Cellular interactions
HLA-G recognizes NK, T and B cells bearing the LILRB1 receptor on their surface [64]. Antigen presenting cells recognize both placental leucocytes and HLA-G + cells, which express LILRB1, and LILRB2 receptors. Also, HLA-G modulates NK cell cytotoxic activity in contact with LILRB1, LILRB2, and KIR2DL4 receptor [86][87][88]. Moreover, LILRB2 receptor in antigen presenting cells and CD8 receptor in CTL cells are recognized by HLA-G [87].
HLA-G*01:05N, -G*01:01 and -G*01:04 alleles World distribution: significance The first confirmed HLA-G null allele was described by Arnaiz-Villena et al. in a Spanish population sample [38]. This HLA-G null allele protein could exist only with a single α1 domain: a single-base deletion induces a shift in the reading frame and a consequent premature stop codon. [3,29,39]. A protective effect against gestational infections has been associated with this allele but also recurrent spontaneous abortions [3,64]. However, the hypothesis that frequent intrauterine infections can maintain high null allele frequencies is discarded, since Mayas and Uros populations, with a weaker health care services in comparison with European ones, do not have this allele. Also, Brazilian and mixed Amerindian populations show similar low frequencies [103]. Middle East Caucasians (Iraqis, Iranians, and Indians from North India) and some African populations (Ghana, Shona, and African Americans) show significantly higher frequencies of this null allele (Fig. 4). HLA-G*01:05N allele DNA sequence indicates that it was probably originated from the HLA-G*01:01 allele: both protein sequences are identical except for a cysteine deletion at codon 129/130 [82]. Moreover, HLA-G* 01:05N allele is in linkage disequilibrium with the HLA-A*30:01-B*13:02 haplotype, which is prevalent in Middle East and some Mediterranean populations. This haplotype may have been introduced in Spain by Muslim invaders in the eighth century AD or long before, when Saharan migrations took place from Saharan Desert to the Mediterranean Basin due to hyperarid climatic conditions beginning about 10,000-6,000 years ago [38,[104][105][106][107]. HLA-G*01:05N "founder effect" could place Middle East as the origin of this allele, because it contains the highest World reported frequencies [62].
As HLA-G is known to play an important role in maternal-fetal tolerance, it is striking how there exist HLA-G*01:05N healthy homozygous mothers capable of giving birth to normal and healthy fetuses. This finding indicates that the HLA-G1 isoform is not crucial for normal pregnancy development [82]. This is also supported by genus Macaca primates which have a normal development during pregnancy and adult life with HLA-G incomplete molecules [108,109]. HLA-G α1 domain could be sufficient for the normal functioning of the HLA-G molecule, so negative evolutionary pressures would not act to eliminate this gene [39] or could be substituted by other HLA class I molecule at the placenta level. Also, HLA-G*01:05N allele may improve the level of immune response against HIV infection [110] or other infections not directly related to pregnancy.
On the other hand, highest frequencies of HLA-G*01:04 allele are found in South Korean, Iranian, and Japanese populations (27.7%, 31.36%, and 45%, respectively) (Fig. 5). Amerindian populations show similar HLA-G*01:04 allele frequencies among them: 10.2% in Uros from Titikaka Lake or 13.1% in Mayans from Guatemala. It is important to point out that HLA-G*01:04 allele frequencies higher than 10% have not been found in Europe neither higher than 13% in South Europe (Spaniards 11%, Portuguese 13%) (Fig. 5). Significant HLA-G differences have not been found, but a trend to lower frequencies in central Europe in comparison with Amerindians is detected (Fig. 6).
Also, higher frequencies of HLA-G*01:01 allele are found in USA South Dakota Hutteritie population, Ghanians, and Germans (79.8%, 83.3%, and 87.4%, respectively). Similar HLA-G*01:04 frequencies are found throughout all Amerindian populations (Fig. 5). Table 1 HLA-G, -E and -F receptors a Structure of this interaction has been defined by X-ray crystallography [95] b Structure of this interaction has been defined by X-ray crystallography [14] c Structure of this interaction has been defined by homology with crystallographic HLA-A2-CD8 and H-2 Kb-CD8 studies [86,96,97] d Bibliography about structure of this interaction has not been found. Only functional assays using monoclonal antibodies have been used to discuss this interaction [11,88,98] e Structure of this interaction has been defined by X-ray crystallography [99] f Structure of this interaction has been defined by homology with crystallographic HLA-E-NKG2A studies [99] g Structure of this interaction has been defined by homology with crystallographic HLA-E-NKG2A studies [99] h Structure of this interaction has been defined by X-ray crystallography [100] i Structure of this interaction has been defined by homology with crystallographic HLA-A2-CD8 and H-2 Kb-CD8 studies [86,96,97] j Bibliography about structure of this interaction has not been found. Only affinity studies have been used to discuss this interaction [91] k Bibliography about structure of this interaction has not been found. Only affinity studies have been used to discuss this interaction [91] l Bibliography about structure of this interaction has not been found. Only functional assays using monoclonal antibodies have been used to discuss this interaction [92] m Bibliography about structure of this interaction has not been found. Only functional assays using monoclonal antibodies have been used to discuss this interaction [92,101] n Bibliography about structure of this interaction has not been found. Only studies of interactions measured by surface plasmon resonance have been used to discuss this interaction [102] o Structure of this interaction has been defined by X-ray crystallography [47] p Bibliography about structure of this interaction has not been found. Only affinity studies have been used to discuss this interaction [

Structure
HLA-E is a heterodimer having an α heavy chain and a light chain (β-2 microglobulin). Heavy chain size is about 45 kDa and it is anchored to the cell membrane. HLA-E gene contains 8 exons. Exon 1 encodes for signal peptide, exons 2 and 3 encode for the α1 and α2 domains (peptide-binding site), exon 4 for the α3 domain, exon 5 for transmembrane domain, and exon 6 for cytoplasmic tail [111]. Exons 7 and 8 are not present in the mature mRNA.

Evolution
Both New World and Old World monkeys MHC-E proteins preserve invariant residues at the tridimensional proteinpresentation valve, like in all other MHC class I molecules from reptilians to humans. Also, the rate of substitutions in peptide-binding site reveals the exixtence of a high evolutionary pressure for stability in this area. MHC-E polymorphism in Macaca mulatta and Macaca fascicularis is restricted to 13 positions in exon 2 (3 synonymous and 10 nonsynonymous variations), 22 in exon 3 (10 synonymous and 12 nonsynonymous substitutions) and at the beginning of exon 4 (2 nonsynonymous variations); in contrast, exon 4 in humans does not show any variation in its sequence. Polymorphism in MHC-E gene of Cercopithecus aethiops is confined only to exon 3 with 1 synonymous and 1 nonsynonymous substitutions [112].
Regarding interspecific studies on MHC-E, an example of trans-specific MHC-E evolution has been found in genus Macaca: Macaca mulatta and Macaca fascicularis share the same MHC-E exon 2 and exon 3 sequence in one allele [112]: both Mamu-Mhc-E-*0101 and Mafa-Mhc-E*04 alleles are identical in exonic 2 and 3 sequences, only differing at the beginning of exon 4 at position 184 [112]. Also, a duplicated MHC-E locus has been found in Macaca mulatta, which may be originated by unequal crosses among different MHC-E homologue locus [113,114]. These duplications have also been reported in other primates class II MHC genes but never before in class I loci [115]. On the other hand, these Macaca mulatta and   [63] a trans-specific evolution of the MHC complex [112,116,117].
It was thought that MHC-G was primordial to other MHC genes in apes, giving rise to other typical MHC antigen presenting alleles, because it is present in New World monkeys (Saguinus oedipus), which are more ancient than Old World monkeys [16,68,118]. However, other data suggest that MHC-G molecules in primates could be non-functional (deletions in genus Macaca) [74] and other MHC proteins, like MHC-E, could do this function instead. It is known that MHC-G molecules in genus Macaca are not able to bind and present peptides and thus being surface expressed, because all individuals bear HLA-G deleted genes, but they may be useful for α1 interactions with cognate receptors [39,74].
With regard to Saguinus oedipus MHC-G allele, it seems to be phylogenetically closer to MHC-E alleles of other species. These analyses were carried out using primary DNA sequences, genetic distances and Neighbor-Joining dendrograms that closely related MHC-G from New World primate (Saguinus oedipus) with MHC-E primary DNA sequences of macaque (Macaca mulatta) and orangutan (Pongo pygmaeus); it is also relevant that genus Macaca lack full MHC-G mRNA transcripts and DNA sequences [3,74,112,118].
It has been shown that HLA-E locus is the most ancient HLA locus in humans, which may support the presence of MHC-E-like molecules in Saguinus oedipus, being the putative primitive MHC gene in primates [119] (Fig. 6).
It seems that selective pressures have occurred to conserve aminoacidic positions in the peptide-binding site of primate MHC-E molecules. It has been also found that MHC-E alleles have suffered trans-specific evolution, duplications, unequal crosses, and substitutions in primates, but it has remained for approximately 40 million years. Indeed, pockets of MHC-E presenting molecules among species, i.e., two human alleles, macaques MHC-E and MHC-E-like molecule in mouse (Qa-1b), have been studied and they all share main aminoacidic anchor portions during million years [120]. Also, human and medium-sized apes (macaques) MHC-E molecules present identical peptides to CD8 + T cells; in man, HLA-E presents leader peptides from class Ia molecules to regulate NK cells [121]. Other studies have also pointed out that MHC-E locus is the most conserved histocompatibility gene in primates, and this ancient evolutionary conservation of MHC-E peptide-binding site structure suggests a crucial relevance in immunological processes [112,122]. (2) HLA-G*01:01 frequencies (green squares) do not clearly differ among World populations [63] 464 Page 10 of 18

DNA transcription and translation
HLA-E transcripts are found in a great variety of tissues, and it is doubtful whether HLA-E molecules reach the surface in normal tissues conditions [123,124]. However, scanty productive allelic changes described are mostly at the T-cell receptor-binding site [125], and it was put forward that HLA-E function may be related to the T-cell repertoire shaping in the thymus or otherwise to presenting a limited peptide repertoire. HLA-E is expressed in the cytoplasm and then on the surface of cytotrophoblast cells but only in the last months of the pregnancy and its expression control is mediated by INF gamma [126][127][128].

Surface molecules
A nonamer peptide derived from residues 3-11 of signal sequences of most classical MHC class I molecules is required for HLA-E cell surface expression [28,49,[129][130][131][132]. This leader peptide is released in the cytosol and then transported by TAP into the lumen of the endoplasmic reticulum, where it binds to HLA-E groove [49,131]. Therefore, HLA-E surface expression allows NK cells to control the expression of a wide range of polymorphic MHC class I molecules through a single receptor. HLA-E surface expression inhibits NK-cell-mediated cytotoxicity [133].

Receptors
ILT2 and ILT4 receptors bind to HLA-E [91]. It also has been shown to interact with other NK cell receptors like NKG2A [87,132], NKG2C, and NKG2E [90]. Moreover, it is known that HLA-E can interact with TCR and CD8 receptors on the surface of CTL cells [48,64,87] (see Table 1).

Cellular interactions
It was detailed above that HLA-E regulates NK cell activity through interaction with LILRB1, LILRB2, NKG2A, NKG2E, and NKG2C molecules : all of them are expressed on the NK cells surface [86,87,134,135]. Also, HLA-E interacts with T CD8 + lymphocytes through TCR and CD8 [87].

HLA-F
Structure HLA-F protein is a ~ 40-41 kDa molecule with HLA class I domains [136]. Due to an alternative splicing process, HLA-F mature mRNA does not contain the exon 7 sequence [137,138], which leads to a modification in the protein, making cytoplasmic tail shorter in comparison to classical HLA class I proteins [137,139] (see Fig. 1).

Evolution
HLA-F orthologous DNA molecules are found in chimpanzee, bonobo, gorilla, and orangutan. Their amino acid sequences and their comparison with other primate MHC-F proteins show that MHC-F is a protein with a class I structure and that the characteristic residues of the peptide-binding region (PBR) are highly conserved in primate MHC-F. Therefore, MHC-F conservation along primate evolution suggests an important role in physiology. Thus, MHC-F protein could function together with MHC-G and MHC-E, in the natural killer (NK) cell activity regulation [140]. HLA-F orthologues have been compared in Pongidae, Macaca and American apes; they present only one pair of active MHC-F genes per chromosome whether they have or not duplicated genes. In addition, a New World (American) monkey, the marmoset, shows 6 orthologous transcripts. However, in all other New World monkeys, Old World ones and humans, MHC-F expansion by duplication has been inactivated to maintain only two parental MHC-F copies per individual irrespective of the number of duplicated copies contained: thus MHC-F gene is under purifying selection [141].
All MHC-F studies in chimpanzee, gorilla, orangutan, Rhesus macaque, and cotton-top tamarin have pointed out a mutation in intron 6 splice site, which drives to the lack of exon 7 in the mature MHC-F protein [69,[137][138][139][140][141][142][143][144]. This common characteristic among these species indicates that this mutation took place before Old World and New World monkeys diverged about 35 million years ago [67,68,140].
MHC-F alleles in human, chimpanzee, bonobo, gorilla, and orangutan lack a six-nucleotide sequence that is present in rhesus macaque and cotton-top tamarin within exon 2. Thus, this six-nucleotide deletion happened in a common ancestor of genera Homo, Pan, Gorilla and Pongo after the separation of rhesus monkeys and New World monkeys' evolutionary pathways. Phylogenetic trees performed show a strong similarity of MHC-F exons 2 and 4 sequences among species: all of them cluster together in a separated tree branch from other class I molecules [140].
Three-dimensional structure of HLA-F is similar to that of the other class I molecules. Also, the little differences observed among sequences of primate species indicate that there must be a strong selective pressure for invariance, except for the Saguinus oedipus (Saoe-F protein), that has a degree of difference of about 15%, while in the comparisons among other primates is under 6% [140].

HLA-F transcription and translation
Molecules of HLA-F are intracellularly expressed in many body cells and tissues; these are peripheral blood lymphocytes (PBL), resting lymphocyte cells (B, T, NK), tonsils, spleen, thymus, kidney, brain, bladder, colon, liver, lymphoblast T-cell leukemia, and tumors. In addition, HLA-F is expressed on fetal extravillous trophoblast cells, which are in close contact with the maternal tissues [113]. HLA-F is expressed both intracellularly and on the surface of cytotrophoblast from the second trimester onwards [91,118,126,145].

HLA-F surface expression
Expression of HLA-F is found on the surface of activated lymphocytes, tumors, HeLa cells, EBV-transformed lymphoblastoid cells, and in some activated monocyte cell lines [89,139]. HLA-F surface expression occurs after immune response activation: HLA-F is found on the surface of stimulated T memory cells but not on circulating regulatory T cells [146].

HLA-F receptors
HLA-F tetramers have been shown to bind LILRB1 and LILRB2 receptors without any peptide binding [47,64,[87][88][89][90][91]94]. HLA-F open-conformed form has also been shown to bind KIR receptors of NK cells, like KIR3DL2 and KIR2DS4 [92,93]. These HLA-F interactions are believed to stabilize other ligand-receptor interactions between trophoblast cells and decidual NK cells during pregnancy. Decidual NK cells play an important role in pregnancy immune regulation; binding to KIR2DS1 has also been shown [93] (see Table 1).

Cellular interactions
It has been shown that HLA-F binds decidual NK cells in the trophoblast during pregnancy. It interacts with active or inactive NK cell activity in the maternal decidua through recognition of KIR3DL2, KIR2DS4 and KIR3DS1; these cells are also recognized through LILRB1 and LILRB2 receptors [87,147]. Moreover, HLA-F recognizes T, B and NK cells which express LILRB1 receptor [64]. HLA-F + decidual leucocytes and antigen presenting cells interact also trough LILRB1 and LILRB2 receptors [64,94].

Nature evolution vs statistical models
MHC was discovered in chicken by B. Briles in 1950 [148]. The first HLA and disease association was described by Amiel in 1967 [149]. Many diseases have been found statistically associated with HLA and MHC classical class I and class II genes. However, today, in 2022, no universally accepted pathogenesis mechanisms have been found to explain classical HLA genes and disease association [64] despite a flood of research on both statistical and in vitro models trying to find out mechanisms and pathogenesis, suggesting pathogenetic proposals which are not yet universally accepted [64].
On the other hand, since Dan Geraghty [7] and Edgardo Carosella groups [150] uncovered HLA-G structure and immune system modulation by this molecule, another flood of HLA-G and disease studies has occurred, particularly in relation to autoimmunity, cancer, and fetal/maternal pathologies. Again, no mechanisms have been clarified up until now.
In the meantime, HLA-E and -F immune suppressive genes have also been studied [128,137]. It is then time to review and study on how Nature modulates the evolution of genes [151] at least in primates, where orthologous genes are well defined. This may give a clue on function and associated pathology of these immune response control molecules, MHC-G, -E and -F. In this article, we have tried to shortly review some of these aspects.
MHC genes for specific, non-specific, and regulatory immunity: extended HLA haplotypes Much debate has occurred, because so many MHC different immune genes go close together in a short chromosome area across species from amniotes to humans during many million years [152][153][154]. This suggests that this set of genes may work together to save individual and species from external injuries, probably microbes, and associated self-recognition pathologies [3,64,155,156]. In this chromosome region lies : a) non-adaptive immunity genes i.e.: C2, C4 and Bf complement factors, tumoral-necrotic factors (TNF) genes, heat shock proteins (HSP) genes, lymphotoxin genes (LTA, LTB) or some zinc finger codifying genes like TRIM40; b) adaptive immunity genes like tapasin (TAP) genes, lymphocyte antigen 6 (Ly6), HLA classical class I (-A, -B, -C), and class II (-DQ, -DR, -DP) genes or MIC genes (MICA, MICB); c) regulatory genes like HLA nonclassical (-G, -E, -F) genes in primates, and others [157,158]. Keeping together a set of certain alleles set of all known immune-related genes may be more advantageous for survival (i.e.: MHC haplotypes rather than single genes) [159] and this may be the reason why all these genes are transmitted conjointly at least from amniotes to humans [3,64,154,158,160]. A search as towhy they are transmitted and work together is worth to follow at this point of MHC/ disease association nihilism. Coevolution of adaptive (i.e.: class I and class II), natural (i.e.: complement), and modulatory (i.e.: HLA-G, HLA-E, HLA-F) genes may point out that studying MHC haplotype/disease association in full or in part may be more fruitful to explain the association of HLA and disease than single-locus allele studies [161,162].

HLA haplotypes and disease association
Thus, the key for understanding HLA association to disease may be studying no single-locus genes but a cluster of neighboring and conjointly transmitted MHC genes (MHC haplotypes). It also would apply to HLA-G extended haplotypes and disease studies [64,155,156]. This approach was already suggested by Roger Dawkins in 1983 [161]: they tried to associate ankylosing spondylitis, rheumatoid arthritis, myasthenia gravis and systemic lupus erythematosus with complotypes (set of C2, Bf and C4 alleles inherited conjointly) and extended HLA haplotypes using different number of neighboring loci alleles. They also related susceptibility to diseases not only with HLA haplotypes but also with retroviruses inserted in the region, which affected expression of MHC genes and also their polymorphism and MHC segment duplication [162]. All or some of these factors within a complotype or a more extended haplotype should be studied to ascertain HLA and disease association. Indeed, this may be technically difficult to study but perhaps more fruitful. More or less long extended HLA haplotypes have been studied with some success in certain diseases; Berger's Disease in 1984 [163], type I diabetes in 1992 [164], and some extended HLA haplotypes were also defined in 1991 [165]. However, relatively few studies have been done up until now; some of them were in microscopic polyangiitis [166], celiac disease [167], kidney disease [168,169], diabetes [170], and psoriatic arthritis [171]. Technical difficulties of this type of study may be in part overcome by nowadays more advanced technologies.

Additional remarks
1. MHC-G complete molecule is lacking in some humans and all primate individuals belonging to genus Macaca.
Other MHC molecules may substitute its function or parts of the molecule may suffice for functionality. 2. Some apes do not have all of the MHC-G soluble isoforms described in man. 3. MHC-E (and not -G) may be the primordial MHC gene in apes, which gave rise to other MHC molecules. 4. A conjoint immune evolution and transmission in a relatively short DNA stretch of MHC, i.e.: immunosuppressive MHC genes (MHC-G, -E, -F), classical presenting molecules and non-adaptive ones (i.e.: C2, C4, Bf) is maintained for a long time from amniotes to human at least, because haplotypes or a specific set of MHC genes/alleles may be necessary for self-maintaining against pathogens and/or other injuries.
Author contributions AA-V and FS-T performed research, planning, design, writing, figures and composition of the manuscript. References and figures revision was performed by IJ, CV-Y and MM-A. Allelism interpretation, database studies and revision of the manuscript were carried out by JP-G. Carmen R-S and EF-C performed manuscript reshaping, references revision, and critical revision of the final manuscript. JMM-V contributed to research, planning, design, and writing of the manuscript. All authors read and approved the final manuscript.
Funding This work was supported by grants from the Spanish Ministry of Science and Universities (PI14/01067, PI18/00720, and PI18/00626) and European FEDER funds.