The self/nonself issue

Abstract

Defining self and nonself is the most compelling challenge in science today, at the basis of the numerous questions that remain unanswered in the immunology-pathology-therapy debate. The generation of the antibody repertoire, the complicated scenario offered by tolerance and autoimmunity, natural auto-antibodies and their relationship to autoimmune diseases, and positive and negative selection are only a few examples of the unresolved immunological questions. In this context, we proposed that sequence similarity to the host proteome modulates antigen peptide recognition and immunogenicity. Using the available proteome assemblies of viruses, bacteria and higher vertebrates, and applying the low-similarity criterion, we are systematically defining the proteomic similarity of B-cell epitopes already validated experimentally. Here, we report further data documenting that a low similarity to the host proteome is the common property that defines the immunological “nonself” nature of antigenic sequences in cancer, autoimmunity, infectious diseases and allergy.

Identifying Epitopes

Antibodies are immunoglobulin proteins that interact with specific areas on the surface of antigen proteins. These areas are referred to as B-cell epitopes. Identifying B-cell epitopes in order to induce a specific antibody response constitutes the unresolved core of immunology. Practically, the identification of epitopes in proteins is the fundamental, preliminary step in designing effective immunotherapy for cancer and infectious diseases as well as autoimmune pathologies.^1–3 As a logical consequence, recent decades have seen determined efforts aimed at identifying and defining antigen epitopes. To understand the molecular determinants characterizing epitopic structures, immunology has used the self/nonself concept.^4–6 For decades, immunologists have based their studies, reasoning, experiments and clinical treatments on the idea that the immune system works by distinguishing between self and nonself.^4,5 However, theoretical and experimental considerations lead to the recognition that there are no known molecular mechanisms that can explain how peptides of self-origin can be discriminated qualitatively from peptides of nonself-origin.⁶ So, for example, how can the hexapeptide VLDVGG, which occurs in two different human proteins, be discriminated from the hexapeptide VLDVGG, which occurs in 380 bacterial, viral, protozoan and other organism proteins? How can the heptapeptide PPPPPPP, which occurs in 625 human proteins, be catalogued as self or nonself and distinguished from the heptapeptide PPPPPPP, which occurs in 12,883 bacterial, viral, protozoan and other organism proteins?

Epitopes: A Problem of Numbers

To answer questions on immunology helpful to immunotherapy, numerous predictive programs and algorithms have been developed for exact epitope identification. Different epitope qualities have been proposed and investigated, including protein hydrophobicity^7–9 protein hydrophilicity,¹⁰ the protrusion index,¹¹ protein flexibility,¹² and protein secondary structure and conformational parameters.^13,14 However, the results of these studies are inconsistent and very little progress in B-cell epitope prediction has been made.^15,16

Generally, the lack of success in B-cell epitope definition is due, mainly, to the difficulties associated with the task. The optimal amino acid (aa) length of a B-cell epitope is five aa,¹⁷ but longer epitopes have been described. In general, linear B-cell epitopes have been described as varying in length, with up to 16 residues reported.^18,19 Considering the 20 naturally occurring amino acids, the potential epitope repertoire ranges from 3,200,000 different linear 5-mers to 655,360,000,000,000 different linear 16-mers. In addition, epitopic amino acid sequences may originate from conformational folding: antigen proteins are long amino acid chains folded into an enormous number of shapes. Searching through all possible foldings to evaluate the working structure-function of an epitopic sequence is a practically impossible task. Finally, epitopic sequences can be derived from post-translational modifications; they can also be site-switched by the epitope spreading phenomena, and can exist as cryptic epitopes. Again, this makes the possible determinant configurations numerically infinite.

The Proteome-Guided Definition of the Immune Object

The proteomic era has provided comprehensive proteome databases for numerous (micro)organism types that can now be analyzed in detail using large-scale proteomics scanning. Using the available proteome repertoire, we advanced and explored the hypothesis that the immunogenicity of peptide sequences is modulated by low similarity to the host's proteome. During the past decade, a series of experimental models involving different disease-associated proteins^20–31 have substantiated and supported the low-similarity hypothesis.

Experimental data from our laboratory have been confirmed by numerous reports in the epitope mapping literature. A systematic analysis of epitopic peptide sequences in cancer, autoimmunity, allergy and infectious diseases indicates that the epitopic boundaries of an amino acid sequence are dictated by the low-similarity hypothesis.^28–30 Only peptide motifs with no/low similarity to the host proteome are epitopic targets in the humoral immune response. Additionally, only amino acid sequences that are scarce (or absent) in host proteins appear to participate in antigen-antibody interactions. In this way, we have validated hundreds of epitopes as low-similarity amino acid sequences.^28–30

This structured survey adds to the tabulation of low-similarity epitopes obtained through proteomic scanning. As a pragmatic example of proteome-guided epitope definition, Table 1 provides a sequence similarity analysis of the most recent data in epitope mapping of disease-associated antigens. The Table shows that amino acid fragment(s) endowed with a low level of similarity to the host proteome represent the common signature of experimentally validated epitopes, regardless of whether they are linear, discontinuous or mimic, independently of their being from microbial antigens, tumor-associated proteins or allergens.

Table 1.

B-cell epitopes are (or contain) sequences with zero/low-similarity to the host proteome

Antigen	Aa position	Epitope sequencea	Matchesb	Proteomec	Ref.
Prostate specific antigen	27–31	GGWEC	0	M	31
Melanin-concentrating hormone receptor 1	254–265	lYWFTLy	0	H	32
Osteopontin	31–37	qlYNKYP	0	M	33
Coagulation factor VIII	Mimotope	npveNMMDRdsq	1	H	34
	Mimotope	qspWQTWFtral	1	H	34
Adenosine A2a receptor	172–178	lfEDVVP	4	M	35
Collagen type IV, α3chain	24–38	FTRHSqttANPSCpe	0	R	36
Receptor tyr kinase-like	518–525	reEFRHEa	1, 2	M	37
CXCR1 chemokine	Mimotope	SFIWDf	3	M	38
	Mimotope	SAMWDf	0	M	38
	Mimotope	TNMWDf	0	M	38
	Mimotope	iTMWDF	1	M	38
	Mimotope	SDWWDf	0	M	38
CXCR2 chemokine	Mimotope	FWDDFw	1	M	38
	Mimotope	LWDDFw	2	M	38
	Mimotope	mWNDFW	0	M	38
	Mimotope	FWLDFw	0	M	38
DNA	Mimotope	DWEYSvwlsn	0	M	39
Bla g 4 allergen	118–152	cpaaanGHVIYvqlrltWRRFHpklgdkeMIQHYT	2, 2, 0	H	40
Blo t 12 isoallergen	30–49	htepddHHEKPtTQCTHeet	0, 0	H	41
	73–92	teeTHHSDdlivHEGGKtyh	0, 2	H	41
	111–130	iicsksgslWYITVmPCSIG	0, 1	H	41
HIV-1 envelope gp41	633–639	iWNNMTw	0	H	42
			0	M	42
HIV-1 envelope gp41	633–642	iwnnMTWMQw	0	H	42
			0	M	42
HIV-1 gp41	675–681	elDKWAS	1	M	43
HIV envelope gp41	670–677	wNWFDItn	0	M	44
Porcine reproductive-respi ratory syndrome virus gp5	152–156	rLYRWR	0	M	45
	169–178	eGHLIDlkrv	4	M	45
	196–200	QWGRL	1	M	45
	196–200	QWGRP	0	M	45
Rabies virus glycoprotein	Mimotope	kRDSTW	2	H	46
	Mimotope	kYLWSK	0	H	46
	Mimotope	kYWLSR	2	H	46
	Mimotope	kYWWSK	0	H	46
	Mimotope	kYAWSR	0	H	46
	Mimotope	kYSMSK	0	H	46
Japanese encephalitis virus NS1 protein	146–150	EHARW	1	M	47
Influenza A H5N1 HA	193–199	qNPTTYi	1	M	48
Influenza A H5N1 HA	121/164d	sWS/YNN	0	H	49
Influenza A fusion peptide HA2 glycopolypeptide		glfgAIAGF	1	M	50
K. pneumoniae adhesin	Mimotope	qktlakSTYMSa	0	M	51
Meningococcal factor H-binding protein	25/57d	DHK/YGn	1	M	52

The analysis was restricted to reports published in 2009 in PubMed.

^aLow similarity 5-mer given in capital letters. Based on the minimum length of an immune unit of 5–6 amino acids,¹⁷ each epitope has been dissected into pentapeptide sequences. The pentapeptides overlapped by four residues, i.e., were shifted by one amino acid. Then, each pentapeptide was analyzed for the number of occurrences in the host proteome using described methodologies.²⁰–³¹ Any occurrence is called a match. A pentapeptide with up to five perfect matches to the host proteome was considered a low-similarity sequence.

^bThe number of matches refers to the low-similarity 5-mer in capital letters.

^cHost proteome: H, human; M, murine; R, rat.

^dDiscontinuous.

Perspectives: Scientific and Clinical Applications

Here, we analyze a robust set of recent experimental reports suggesting that a low level of sequence similarity to the host proteome defines the B-cell epitope pool in the humoral immune response. The data add to and support experimental results from our laboratory^21–27,31 and reported meta-analyses.^28–30

Theoretically, proteomic similarity analyses might elucidate the regulatory mechanisms/factors that dictate peptide immunogenicity assessment. From a clinical perspective, low-similarity peptides may have strong repercussions affecting the rational development of peptide-based treatments in cancer, infection and autoimmunity. De facto, the most attractive feature of the similarity concept is that it appears to guarantee the highest specificity and lowest cross-reactivity when designing effective, safe and theoretically infallible immunotherapeutic tools. The Ehrlichian idea of therapeutic agents equipped with high affinity to the causative agent, and efficacy at concentrations harmless to the patient, appears feasible.

Scientifically, the solution to what has been called the top-ranking mystery in science appears to be at hand, i.e., the mechanism of self versus nonself recognition in the immune system.⁵³ In fact, the low-similarity hypothesis might represent a quantum leap forward in our understanding of the rules dictating the immunological antigen-antibody epitope-paratope interaction at the peptide level.