**1. Introduction**

ERAP1 and ERAP2 (endoplasmic reticulum aminopeptidases 1 and 2) are two IFNγ- and TNF α-inducible, ubiquitously-expressed human enzymes, which belong to the M1 family of zinc aminopeptidases [1]. In the endoplasmic reticulum (ER), ERAPs shape the antigenic repertoire by trimming the N-terminus of precursor peptides previously generated in the cytoplasm by the proteasome. In this way, ERAPs generate optimal-length peptides for loading onto MHC class I groove to be presented to CD8+ T lymphocytes [2,3]. Despite maintaining marked di fferences in their enzymatic specificity these two enzymes can act together in a concerted way, through the formation of homo- or heterodimers, thus allowing the generation of a variegated and more immunogenic antigenic

repertoire [4]. In particular, ERAP1–ERAP2 heterodimer generation has been demonstrated to improve the shaping of peptides suitable for MHC class I molecule binding [5].

ERAPs are encoded by two genes, sharing ~49% sequence homology and situated on chromosome 5q15 in opposite directions, which are highly polymorphic [6]. Since their leading role in the antigen processing pathway, several studies have investigated any potential link between ERAP polymorphic variants and alterations in their functioning, which could result in MHC-I-associated disorder onset [2,7,8] as well as into variations in susceptibility/progression to microbial infections [9].

As for ERAP2, the most relevant single nucleotide polymorphism (SNP) is the non-coding rs2248374 (A/G) which identifies two haplotypes, hereafter referred to as HapA (A allele for rs2248374) and HapB (G allele for rs2248374). In HapB, the G allele for this SNP primes the transcription of a spliced ERAP2 variant (ERAP2/Iso2), presenting an extended exon 10 (56 extra nucleotides) and two in-frame TAG stop codons, which in turn lead to its nonsense-mediated decay (NMD) [10]. Conversely, HapA is translated into a 965-amino-acid protein and is associated with Crohn's disease [11], HLA-A29-associated birdshot uveitis [12], ankylosing spondylitis [13,14] and juvenile idiopathic arthritis [15], as well as natural resistance to HIV infection [9,16,17]. Since these variants are maintained by a balanced selection to a frequency of approximately 50% (HapB: 53% and HapA: 47%), nearly 25% of the population fails to express the ERAP2 protein [18]. This observation raises a logic question: in which peculiar setting does balancing selection operate to conserve the apparently loss-of-function HapB and the disease-causing HapA in the human population? Quite recently, Ye and co-workers provided an exhaustive explanation to this apparent paradox [19]. Indeed, for the first time, they documented the transcription of two novel short isoforms (ERAP2/Iso3, ERAP2/Iso4) from flu-infected monocyte-derived dendritic cells isolated from homozygous HapB-carrying subjects [19]. The two short isoforms are transcribed from HapB and di ffer from the full-length one—ERAP2/Iso1, transcribed from HapA—since their transcription begins in correspondence of exon 9 and undergoes alternative splicing of an extended exon 10. Besides, they diverge from each other by alternative splicing at a secondary splice site at exon 15 (Figure 1). Of note, while ERAP2/Iso4 is predicted to harbor a premature termination codon that could lead to NMD, ERAP2/Iso3, is expected to be translated into a protein [19]. Such protein misses the catalytic domain [19], but could still critically contribute to profile the cellular immune-peptidome as it preserves the capacity to dimerize with ERAP1 and possibly ERAP2-wild type (wt) [20].

**Figure 1.** Genetics of ERAP2 isoforms regulation. Structures of transcripts derived from each ERAP2 isoform are represented. Start and stop codons for each isoform are reported.

Based on these premises, the aim of our study was to investigate if the transcription of ERAP2/Iso3 is either flu-specific or if it can be induced by other kind of stimuli, such as other viruses, bacteria, or inflammatory triggers. Given the foremost role played by ERAPs in the field of both acquired and innate immunity, the characterization of the di fferent isoforms produced in a particular pathological setting, such as the one caused by microbial infections, could lead to the identification of new molecular targets to be exploited in the setting up of innovative therapeutically or vaccinal approaches.
