© 1999 Robert A. Freitas Jr. All Rights Reserved.

Robert A. Freitas Jr., Nanomedicine, Volume I: Basic Capabilities, Landes Bioscience, Georgetown, TX, 1999

8.5.2.2 Identification of Cell Type

How many different cell types are there in an adult human body? The usual estimate based on histological studies is that there are ~200 distinct kinds of cells that show alternate structures and functions.^312,531,866 These represent discrete categories of cell types of markedly different character, not arbitrary subdivisions along a morphological continuum. Traditional classification is based on microscopic shape and structure, and on crude chemical nature (e.g., affinity for various stains), but newer immunological techniques have revealed, for instance, that there are more than 10 distinct types of lymphocytes. Pharmacological and physiological tests have revealed many different varieties of smooth muscle cells -- for example, uterine wall smooth muscle cells are highly sensitive to estrogen and (in late pregnancy) oxytocin, while gut wall smooth muscle cells are not. Appendix C presents a catalog of the cells of the human body, slightly modified from an original compilation by Alberts et al.⁵³¹ The catalog is organized by cellular function and omits subdivisions of smooth muscle cells, neuron classes in the CNS, various related connective tissue and fibroblast types, and intermediate stages of maturing cells such as keratinocytes (only the stem cell and differentiated cell types are given). Otherwise, the catalog is said to represent an exhaustive listing of the ~219 cell varieties found in the adult human phenotype. (Complexity theory and phylogenetic comparisons suggest that the maximum number of cell types N_cell ~ N_gene^1/2 = 370 cell types for humans with N_gene ~ 10⁵ genes).^1266,766

Medical nanorobots can probably distinguish all of these cell types by surface chemical assay using chemosensor pads (Section 4.2.8). Antigenic specificities exist for species (xenotype), organ, tissue or cell type for almost all cells -- possibly involving as many as ~10⁴ distinct antigens. A comprehensive analysis of the cytoimmunography of all cell types is beyond the scope of this book, but a few examples of cell type-specific antigenic markers can be given to illustrate the tremendous power of this approach.

In the case of red blood cells, antigens in the Rh, Kell, Duffy, and Kidd blood group systems are found exclusively on the plasma membranes of erythrocytes and have not been detected on platelets, lymphocytes, granulocytes, in plasma, or in other body secretions such as saliva, milk, or amniotic fluid.⁹⁶⁰ Thus detection of any member of this four-antigen set establishes a unique marker for red cell identification. MNSs and Lutheran antigens are also limited to erythrocytes with two exceptions: GPA glycoprotein (MN activity) also found on renal capillary endothelium,⁹⁶¹ and Lu^b-like glycoprotein which appears on kidney endothelial cells and liver hepatocytes.⁹⁶² In contrast, ABH antigens are found on many non-RBC tissue cells such as kidney and salivary glands.⁹⁵⁵ In young embryos ABH can be found on all endothelial and epithelial cells except those of the central nervous system.⁹⁶³ ABH, Lewis, I and P blood group antigens are found on platelets and lymphocytes, at least in part due to adsorption from the plasma onto the cell membrane. Granulocytes have I antigen but no ABH.⁹⁶⁰

Platelets also express platelet-specific alloantigens on their plasma membranes, in addition to the HLA antigens they already share with body tissue cells. Currently there are five recognized human platelet alloantigen (HPA) systems that have been defined at the molecular level (Table 8.15). The phenotype frequencies given are for the Caucasian population; frequencies in African and Asian populations may vary substantially. For instance, HPA-1b is expressed on the platelets of 28% of Caucasians but only 4% of the Japanese population.⁹⁶⁴ A medical nanorobot that detects a representative antigen from any of these five HPA antigen groups on an encountered surface can be certain that it has found a platelet.

Lymphocytes with a particular functional activity can be distinguished by various differentiation markers displayed on their cell surfaces. For example, all mature T cells express a set of polypeptide chains called the CD3 complex. Helper T cells also express the CD4 glycoprotein, whereas cytotoxic and suppressor T cells express a marker called CD8.⁹³⁹ Thus a nanorobot that detects the phenotype CD3⁺CD4⁺CD8^- has positively identified a helper T cell, whereas the detection of CD3⁺CD4^-CD8⁺ uniquely identifies a cytotoxic or suppressor T cell. All B lymphocytes express immunoglobulins (their antigen receptors, or Ig) on their surface and can be distinguished from T cells on that basis, e.g., as Ig⁺ MHC Class II⁺.⁹³⁹

Lymphocyte surfaces also display distinct markers representing specific gene products that are expressed only at characteristic stages of cell differentiation. For example, Stage I Progenitor B cells display CD34⁺PhiL^-CD19^-; Stage II, CD34⁺PhiL⁺CD19^-; Stage III, CD34⁺PhiL⁺CD19⁺; and finally CD34^-PhiL⁺CD19⁺ at the Precursor B stage.⁹⁶⁵

There are neutrophil-specific antigens and various receptor-specific immunoglobulin binding specificities for leukocytes. For instance, monocyte FcRI receptors display the measured binding specificity IgG1⁺⁺⁺IgG2^-IgG3⁺⁺⁺ IgG4⁺, monocyte FcRIII receptors have IgG1⁺⁺IgG2^-IgG3⁺⁺IgG4^- , and FcRII receptors on neutrophils and eosinophils show IgG1⁺⁺⁺IgG2⁺IgG3⁺⁺⁺ IgG4⁺.⁹⁵⁵ Neutrophils also have b-glucan receptors on their surfaces.¹⁴⁰³

Tissue cells display specific sets of distinguishing markers on their surfaces as well. Thyroid microsomal-microvillous antigen is unique to the thyroid gland.⁹⁵⁵ Glial fibrillary acidic protein (GFAP) is an immunocytochemical marker of astrocytes,⁹⁴⁷ and syntaxin 1A and 1B are phosphoproteins found only in the plasma membrane of neuronal cells.¹⁰⁷⁹ Alpha-fodrin is an organ-specific autoantigenic marker of salivary gland cells.⁹⁶⁸ Fertilin, a member of the ADAM family, is found on the plasma membrane of mammalian sperm cells.²¹⁴³ Hepatocytes display the phenotypic markers ALB⁺⁺⁺GGT^-CK19^- along with connexin 32, transferrin, and major urinary protein (MUP), while biliary cells display the markers AFP^-GGT⁺⁺⁺CK19⁺⁺⁺ plus BD.1 antigen, alkaline phosphatase, and DPP4.⁹⁶⁶ A family of 100-kilodalton plasma membrane guanosine triphosphatases implicated in clathrin-coated vesicle (Section 8.5.3.7) transport include dynamin I (expressed exclusively in neurons), dynamin II (found in all tissues), and dynamin III (restricted to the testes, brain, and lungs), each with at least four distinct isoforms; dynamin II also exhibits intracellular localization in the trans-Golgi network.¹¹⁹³ Table 8.16 lists numerous unique (to one or the other cell type) and shared antigenic markers of hepatopoietic (e.g., hepatoblast) and hemopoietic (e.g., erythroid progenitor) cells. There are many tumor-specific antigens, a fact exploited in conventional peptide-guided chemotherapy.¹¹⁴³

Bacterial membranes are also quite distinctive, including such obvious markers as the family of outer-membrane trimeric channel proteins called porins in Gram-negative bacteria like E. coli^1045,1053 and other surface proteins such as Staphylococcal protein A¹⁰⁴⁶ or endotoxin (lipopolysaccharide or LPS), a variable-size carbohydrate chain that is the major antigen of the outer membrane of Gram-negative bacteria. Mycobacteria contain mycolic acid in their cell walls.²¹³⁴ In addition, only bacteria employ right-handed amino acids in their cellular coats, which helps them resist attack by digestive enzymes in the stomach and other organisms. Peptidoglycans, the main structural component of bacterial walls, are cross-linked with peptide bridges that contain several unusual nonprotein amino acids and D-enantiomeric forms of Ala, Glu, and Asp.¹⁷¹⁸ D-alanine is the most abundant D-amino acid found in most peptidoglycans and the only one that is universally incorporated.¹⁷¹⁹

At least four major families of cell-specific cell adhesion molecules had been identified by 1998 -- the immunoglobulin (Ig) superfamily (including N-CAM and ICAM-1), the integrin superfamily, the cadherin family and the selectin family (see below).

Integrins are ~200 kilodalton cell surface adhesion receptors expressed on a wide variety of cells, with most cells expressing several integrins. Most integrins, which mediate cellular connection to the extracellular matrix, are involved in attachments to the cytoskeletal substratum. Cell-type-specific examples include platelet-specific integrin (a_IIbb₃), leukocyte-specific b₂ integrins, late-activation (a_Lb₂) lymphocyte antigens, retinal ganglion axon integrin (a₆b₁) and keratinocyte integrin (a₅b₁).⁹⁷⁵ At least 20 different heterodimer integrin receptors were known in 1998.

The cadherin molecular family of 723-748-residue transmembrane proteins provides yet another avenue of cell-cell adhesion that is cell-specific.⁹⁷⁷ Cadherins are linked to the cytoskeleton. The classical cadherins include E- (epithelial), N- (neural or A-CAM), and P- (placental) cadherin, but in 1998 at least 12 different members of the family were known.⁹⁸⁰ They are concentrated (though not exclusively found) at cell-cell junctions on the cell surface and appear to be crucial for maintaining multicellular architecture. Cells adhere preferentially to other cells that express the identical cadherin type. Liver hepatocytes express only E-; mesenchymal lung cells, optic axons and neuroepithelial cells express only N-; epithelial lung cells express both E- and P-cadherins. Members of the cadherin family also are distributed in different spatiotemporal patterns in embryos, with the expression of cadherin types changing dynamically as the cells differentiate.⁹⁷⁷

Carbohydrates are crucial in cell recognition. All cells have a thin sugar coating (the glycocalyx; Section 8.5.3.2) consisting of glycoproteins and glycolipids, of which ~3000 different motifs had been identified by 1998. The repertoire of carbohydrate cell surface structures changes characteristically as the cell develops, differentiates, or sickens. For example⁴¹⁵:

1. The array of carbohydrates on cancer cells is strikingly different from that on normal cells.

2. A unique trisaccharide (SSEA-1 or Le^x) appears on the surfaces of cells of the developing embryo exactly at the 8- to 16-cell stage when the embryo compacts from a group of loose cells into a smooth ball.

3. Bacterial pathogens often use cell-specific sugars⁹⁸¹ to guide them to their preferred targets -- E. coli are abundant in tissues surrounding the ureters leading from kidneys to bladder but are rarely found in the upper respiratory tract; group A streptococci, which colonize only the upper respiratory tract and skin, rarely cause urinary tract infections; the gonorrhea organism Neisseria gonorrhoeae adheres only to cells of the genital and oral epithelia but not to cells of other organs.⁴¹⁵ These bacterial carbohydrate specificities (e.g., carbohydrate-specific bacterial adhesins, lectins, and glycoconjugates) have been at least partially catalogued.^{981,3356-3370} and molecular dynamics studies have begun.^3353-3355

Carbohydrate motifs are in theory more combinatorically diverse than nucleotide or protein-based structures. While nucleotides and amino acids can interconnect in only one way, the monosaccharide units in oligosaccharides and polysaccharides can attach at multiple points. Thus two amino acids can make only two distinct dipeptides, but two identical monosaccharides can bond to form 11 different disaccharides because each monosaccharide has 6 carbons, giving each unit 6 different attachment points for a total of 6 + 5 = 11 possible combinations. Four different nucleotides can make only 24 distinct tetranucleotides, but four different monosaccharides can make 35,560 unique tetrasaccharides, including many with branching structures.⁴¹⁵ A single hexasaccharide can make ~10¹² distinct structures, vs. only 6.4 x 10⁷ structures for a hexapeptide; a 9-mer carbohydrate has a mole of isomers.³¹²²

The coded sugar coating of red cells has already been described (Fig. 8.35). As another example, the CD44 family of transmembrane glycoproteins are 80-95 kilodalton cell adhesion receptors that mediate ECM binding, cell migration and lymphocyte homing. CD44 antigen shows a wide variety of cell-specific and tissue-specific glycosylation patterns, with each cell type decorating the CD44 core protein with its own unique array of carbohydrate structures.^972,973 Distinct CD44 cell surface molecules have been found in lymphocytes, macrophages, fibroblasts, epithelial cells, and keratinocytes. CD44 expression in the nervous system is restricted to the white matter (including astrocytes and glial cells) in healthy young people, but appears in gray matter accompanying age or disease.⁹⁷² A few tissues are CD44 negative, including liver hepatocytes, kidney tubular epithelium, cardiac muscle, the testes, and portions of the skin.⁹⁷²

The selectin family of ~50 kilodalton cell adhesion receptor glycoprotein molecules^976,978 can recognize diverse cellsurface antigen carbohydrates and help localize leukocytes to regions of inflammation (leukocyte trafficking). Selectins are not attached to the cytoskeleton.⁹⁸⁰ Leukocytes display L-selectin, platelets display P-selectin, and endothelial cells display E-selectin (as well as L and P) receptors. Cell-specific molecules recognized by selectins include tumor mucin oligosaccharides (recognized by L, P, and E), brain glycolipids (P and L), neutrophil glycoproteins (E and P), leukocyte sialoglycoproteins (E and P), and endothelial proteoglycans (P and L).⁹⁷⁶ The related MEL-14 glycoprotein homing receptor family allows lymphocyte homing to specific lymphatic tissues coded with "vascular addressin" -- cell-specific surface antigens found on cells in the intestinal Peyer's patches, the mesenteric lymph nodes, lung-associated lymph nodes, synovial cells and lactating breast endothelium. Homing receptors also allow some lymphocytes to distinguish between colon and jejunum.^937,974 Selectin-related interactions, along with chemoattractant receptors and with integrin-Ig, regulate leukocyte extravasation (Section 9.4.4.1) in series, establishing a three-digit "area code" for cell localization in the body.¹⁴⁹⁵

Viral capsid proteins are readily recognized, permitting identification of many virus species.^3371-3375 Enveloped viruses, which have acquired a lipid membrane coating borrowed from the host cell upon release,^1969,3375 will be more difficult to detect by simple surface marker sensing. A lipid-penetrating sensor or a host-cell "decoy" sensor (activating membrane fusion protein release (Section 9.4.5.4) by the virus, thus revealing its true identity) may be more useful.

Finally, cells may be typed according to their indigenous transmembrane cytoskeleton-related proteins. For example, erythrocyte membranes contain glycophorin C (~25 kilodaltons, ~3000 molecules/micron²) and band 3 ion exchanger (90-100 kilodaltons, ~10,000 molecules/micron²);^980,3655 platelet membranes incorporate the GP Ib-IX glycoprotein complex (186 kilodaltons); cell membrane extensions in neutrophils require the transmembrane protein ponticulin (17 kilodaltons); and striated muscle cell membranes contain a specific laminin-binding glycoprotein (156 kilodaltons) at the outermost part of the transmembrane dystrophin-glycoprotein complex.⁹⁸⁰ There are also a variety of carbohydrate-binding proteins (lectins) that appear frequently on cell surfaces, and can distinguish different monosaccharides and oligosaccharides.⁴¹⁵ Cell-specific lectins include the galactose (asialoglycoprotein)-binding and fucose-binding lectins of hepatocytes, the mannosyl-6-phosphate (M6P) lectin of fibroblasts, the mannosyl-N-acetylglucosamine-binding lectin of alveolar macrophages, the galabiose-binding lectins of uroepithelial cells, and several galactose-binding lectins in heart, brain and lung.^{415,970,981,3361}

Each cell expresses a different set of genes from the genome, and each gene normally represents a different protein (although in some cases alternate splicing can generate several proteins from the same gene; Chapter 20). Thus it seems plausible to conclude that each different cell type uses a unique constellation of proteins in its construction which can be detected as a set by nanorobot chemosensors, thus unambiguously identifying the cell type. (Of course, possibly up to ~50-70% of the proteins in different cells may be the same, serving common "housekeeping" functions.) If there are ~400 cell types (including all varieties of smooth muscle cells, neuron cells, and other cytochemically distinct cell subcategories not fully enumerated in Appendix C), then every cell type in the catalog may be uniquely identified using as few as ~log₂(400) ~ 9 binary antigenic markers. Given that Nature has employed ~580 markers among the HLA specificities and ~254 markers in the blood group systems to distinguish self from nonself, it is clear that biological systems may employ considerable multifunctionality and redundancy. By 1998 a complete catalog of all cell-specific antigens had not yet been compiled. However, the completion of sequencing of ~90% of the human genome by the spring of 2000³¹⁸⁶ (with the complete version finished by 2002-2003^3186,3187), and foreseeable advances in DNA chip technology (used to sample gene expression patterns), should allow the rapid determination of unique sets of cell-specific antigens by circa 2002-2005.

In the worst case, 9 unique antigenic markers would be required for each of at most ~400 cell types, giving a maximum of 3600 antigenic markers needed to positively identify all cell types. From Eqn. 8.5, a ~(300 nm)² chemotactic pad with N_spec = 3600 implies a maximum cell typing measurement time t_meas ~ 400 sec in mapping mode. If a more plausible working set of ~300 key antigenic markers will suffice, then t_meas ~ 3 sec in mapping mode. On the other hand, a nanorobot searching for a particular set of N_spec = 9 binary antigenic markers (e.g., seeking only liver cells, rejecting all nonliver cells) requires just t_meas ~ 0.002 sec to make the cell-type determination.

Last updated on 20 February 2003