The PIN-FORMED (PIN) proteins are a plant-specific family of transmembrane proteins that transport the plant signal molecule (phytohormone) auxin as their substrate. Although the limited available data suggest that auxin as a signaling molecule is of an ancient origin in the Plantae supergroup 12, the representatives of the PIN family have been found only in the genomes of land plants (Figure 1). In land plants, the PIN proteins act as key regulators in multiple developmental events ranging from embryogenesis through morphogenesis and organogenesis to growth responses to environmental stimuli. Most of the PIN proteins characterized are located in the plasma membrane and are restricted to particular faces of the cell; they can therefore mediate directional auxin fluxes within tissues and generate auxin maxima and gradients that influence development 34.

The first PIN family members identified and associated with auxin transport were described in the model plant Arabidopsis thaliana. The significance and function of AtPIN1 was discovered through the phenotype generated by the loss-of-function mutation in the gene: mutant plants fail to develop floral organs properly and generate naked, pin-like inflorescences, which gave the name PIN-FORMED (PIN) to the family 56. At the same time, several groups identified the homologous protein AtPIN2 under different names on the basis of a strong root agravitropic phenotype of the loss-of-function mutant. Independently identified mutant alleles of PIN2 were pin2, ethylene insensitive root1 (eir1), agravitropic1 (agr1), and wavy6 (wav6) 78910. Altogether, Arabidopsis has eight annotated PIN genes, of which six have been functionally characterized up to now: PIN1 6, PIN2 78910, PIN3 11, PIN4 12, PIN5 13, and PIN7 14. PIN6 and PIN8 are still awaiting characterization.

The eight Arabidopsis PIN genes generally can be divided into two broad subfamilies. The prominent feature of the larger subfamily is the distinct central hydrophilic loop separating two hydrophobic domains of about five transmembrane regions each (Figure 2). This subfamily of 'long' PINs encompasses all members of the family that are defined as auxin-efflux carriers localized at the plasma membranes (PIN1-PIN4 and PIN7 as well as their homologs from seed plants - called the canonical PINs) 1516. In addition, we include PIN6 also as a member of the long PIN subfamily on the basis of the high sequence similarity in the transmembrane regions and only partial reduction of the hydrophilic loop. The hydrophilic loop is the most divergent part of PIN proteins. On the basis of the sequence of this loop, the long PINs are divided into seven groups (groups 1, 2, and 4-8), such that members of the same group share significant homology in their hydrophilic loops. Two of these groups are represented in every sequenced seed plant genome: AtPIN2- and AtPIN3-related genes form groups 4 and 7, respectively, as shown in Figure 3. Groups 5 (which is divided into subgroups 5a and 5b) and 6 form one monophyletic clade (Figure 3). Subgroup 5a contains the archetypal AtPIN1 and its other dicot-specific orthologs, while subgroups 5b and group 6 contain only monocot-specific sequences. Even though the groups differ markedly in terms of the hydrophilic loop, they may be classified as orthologous on the basis of sequence similarity in the transmembrane regions. Indeed, experimental observations show that ZmPIN1a (maize) 17 and OsPIN1b (rice) 18 in monocots display expression patterns and have developmental roles that are analogous with the expression and developmental role of AtPIN1 in dicots. The last group of sequences in seed plants (group 8) is related to AtPIN6 and contains genes that differ considerably from the other long PIN groups (the canonical PINs). The central hydrophilic loop is markedly reduced and recent data suggest that AtPIN6 is predominantly localized in the endoplasmic reticulum membrane 13.

With the possible exception of the PIN6-related proteins, the general function of all long PINs from seed plants is to transport auxin out of the cell. The groups differ in the regulation of their expression, localization and activity rather than in the auxin-transport function itself. It has been shown, for example, that AtPIN1 and AtPIN2, which are distinct representatives of the long PINs, can functionally replace each other in planta when expressed in the same cells and localized at the same side of the cell 1619.

The second major PIN gene subfamily encodes proteins with the central hydrophilic loop virtually absent ('short' PINs) and comprises AtPIN5 and AtPIN8. Sequence diversification within the subfamily of short PINs tends to be higher than between the long PINs. From this subfamily, only AtPIN5 has been characterized so far 13, and reveals a striking difference from the canonical long PINs in its subcellular localization and thus in its physiological function (see below). The short PINs appear to localize to a large extent to the endoplasmic reticulum, and although they presumably act as auxin transporters, they do not directly facilitate auxin transport between cells but mediate intracellular auxin compartmentalization and homeostasis 13.

The precise origin of PIN proteins in the evolutionary history of plants is not known. The basal split of the Viridiplantae - that is, the separation of the Streptophyta (the clade containing land plants (Embryophyta) and some green algae) from the Chlorophyta (representing the majority of green algae) - probably occurred some 725-1,200 million years ago 20 (Figure 1). All green algae with genomes sequenced so far (Chlamydomonas, Ostreococcus and Micromonas) belong to the clade Chlorophyta and none of these organisms contains a PIN gene. On the other hand, sequence data from the most primitive land plants available - the moss Physcomitrella patens and the club moss Selaginella moellendorffii - have revealed the presence of PIN genes of groups 1 and 2, both belonging to the long PIN subfamily. Nonetheless, to assess the evolutionary origin of PIN proteins more precisely, the genomic data from algae more closely related to land plants (that is, from the Streptophyta) and also from the liverworts, land plants even more ancient than the club mosses, is needed. Interestingly, the P. patens and S. oellendorffii PINs do not cluster with PINs of seed plants or with each other (Figure 3, groups 1 and 2), suggesting separate evolutionary establishment of PIN families in each of the lineages. The only exception is P. patens PpPIND (accession number XP_001765763), which is in the same group as AtPIN6. However, its intron sequences suggest the possibility of horizontal transfer of this gene from monocots 13.

The intron/exon organization of PIN genes is highly conserved. With a few exceptions, the genes are composed of six exons. The first corresponds to the amino-terminal transmembrane segment and most of the central hydrophilic loop. The second exon spans the rest of the variable region of the loop and also the first part of the carboxy-terminal transmembrane domain. It is followed by four small conserved exons coding for the rest of the second transmembrane segment (Figure 4). Several exceptions to this organization exist only in short PINs, and in long PINs related to AtPIN2 (group 4 PINs), where some orthologs display a split of the first canonical exon into two exons.

The predicted structure of canonical long PIN proteins is similar to the structures of secondary transporters - that is, membrane transport proteins that use the electrochemical gradient across the membrane, rather than ATP hydrolysis, to power transport. The PIN proteins have two hydrophobic domains (each with five transmembrane helices) that are separated by a hydrophilic domain with a presumably cytoplasmic orientation. This predicted structure is based only on bioinformatic analyses of the sequen available and has not been verified experimentally. The hydrophobic domains of PIN proteins are highly conserved in sequence, mainly in the transmembrane helices, which tolerate no insertions or deletions; the loops between the transmembrane helices within the hydrophobic domains exhibit much greater variability both in size and sequence. The hydrophilic domains of PIN proteins from the same group (Figure 3) are very similar in sequence, but there is only limited sequence similarity between hydrophilic domains of PINs from different groups.

There is a substantial difference in the sequence variability of the hydrophobic domains between short and long PINs. The hydrophobic domains of long PINs contain positions that have the same amino acid in all available sequences - that is, they are invariant - but not all of these positions are invariant in the short PINs. However, there are no amino-acid positions that are invariant in short PINs but not in the long PINs (Figure 2). This indicates that the positions that are invariant only in long PINs must be crucial for some important function of long PINs that has not been retained in short PINs.

Two motifs important for intracellular trafficking of PIN proteins can be predicted. One comprises two diacidic motifs presumably important for trafficking of proteins from the endoplasmic reticulum that are located in the amino-terminal part of the hydrophilic domain of all long PINs. The other is a tyrosine-based internalization motif present in all PINs that is important for recruitment of the protein into clathrin-dependent vesicles. The importance of these residues for PIN action, however, remains to be demonstrated.

Out of the eight PIN proteins in Arabidopsis, the canonical long PINs are already well characterized and their developmental roles in generating intercellular auxin distribution patterns have been demonstrated 55. On the other hand, the existence of auxin transport into the endoplasmic reticulum and its role in regulating auxin homeostasis is a novel and unexpected finding and there is still lot of work needed to elucidate the details and physiological importance of this activity. From the evolutionary point of view, it would be interesting to know which function of PINs is the older: the plasma-membrane-based intercellular auxin transport by long PINs or the endoplasmic-reticulum-based control of intracellular auxin homeostasis by short PINs? The most ancient PIN proteins currently known, from mosses, are localized to the endoplasmic reticulum, which suggests that intracellular function is evolutionarily ancestral, but this remains to be experimentally verified. The other obvious open questions relate to experimental information on PIN protein structure and membrane topology. This, as well as more detailed biochemical characterization of PIN-driven auxin transport is still largely lacking.