Sperm maturation in epididymis requires successive sperm surface modifications that promote the development of capacities for directional swimming and egg fertilization [1]. Although numerous secreted epididymal proteins that interact with spermatozoa have been identified and characterized, our understanding of sperm maturation is far from complete [2, 3]. Possible lipocalin involvement in sperm maturation was highlighted by recent reports describing the spermatozoon-association of Lcn5 also known as mouse epididymal protein 10 (Mep10) and epididymal retinoic acid binding protein (Erabp) [4], bull prostaglandin D synthase (PTGDS) [5], rat androgen-regulated secretory protein B [6], lizard lipocalin sperm binding protein [7] and mouse Lcn2/24p3 [8]. Other lipocalins are implicated in reproductive processes including mouse Lcn8 also known as epididymal protein 17 (mEP17) [9], glycodelin also known as pregnancy associated endometrial protein (PAEP) [10, 11], aphrodisin [12] and the rodent major urinary proteins (Mups). Mups are pheromone carriers that mediate a range of reproductive effects including accelerated puberty [13], behavioral responses [14] and pregnancy block [15].

The lipocalin genes have duplicated repeatedly during evolution expanding to over 110 total known genes [16], more than 15 reported in human. Expression of this diverse family in all the major taxonomic groups from prokaryotes to plants, invertebrates and vertebrates [17] suggests that their activities are fundamental to evolutionary success. Lipocalins function in a broad range of systems [18] including taste and odor chemoreception and transport, coloration, immune modulation, prostaglandin D synthesis, and several aspects of cell regulation including receptor-mediated induction of apoptosis [19]. In contrast to divergent amino acid sequences and differing functions, the β-strands of different lipocalins fold to form remarkably similar 3-dimensional basket-like structures. The β-barrel is closed at one end by amino acid side chains and open at the opposite end permitting specific entry and binding of retinoids, steroids, terpenoids, fatty acids and a range of aromatic and aliphatic compounds [18]. Ligands are often tightly bound with equilibrium dissociation constants in the 1–100 nanomolar range [20–22] and some in the 1–100 micomolar range [23, 24]. The rodent epididymal lipocalins, Lcn5 and Lcn8 bind retinoic acid, a vital regulatory factor in the male reproductive tract [25]. Lcn5 and 8 may transport retinoic acid from the proximal regions of the epididymis to either spermatozoa or epithelium at more distal locations [26]. Rodent lipocalins are abbreviated Lcn and human lipocalins are abbreviated in uppercase, LCN.

To characterize the novel lipocalin LCN6, we analyzed its sequence, expression in epididymis and localization on spermatozoa. We show that LCN6 expression is highly epididymis-specific, but appears to be regulated little if at all by androgen but may be regulated by testis factors in corpus. We demonstrate LCN6 location on spermatozoa, consistent with a role in fertility.

To investigate novel proteins involved in sperm maturation, the expressed sequence tag (EST) database of Human Genome Sciences Inc, Rockville, MD was searched for epididymis-specific cDNA clones. From more than 130 clones obtained, a cDNA encoding a novel lipocalin, LCN6 was selected for analysis in part because of its close relationship to two well studied rodent epididymal lipocalins, Lcn5 and Lcn8. The human LCN6 gene corresponds to the 5' half of Unigene cluster Hs.98132, LOC158062 on chromosome 9q34 next to the human orthologs of Lcn5 and Lcn8, in a region rich in lipocalin genes (Fig. 1). The Locus158062 and Unigene cluster information are not shown in Fig. 1, but are available at the National Center for Biotechnology Information (NCBI) website http://www.ncbi.nlm.nih.gov/mapview/maps.cgi?. The human LCN6 sequence is based on more than 10 clones we isolated during library screening.

The relative positions of LCN6 and representative related genes are indicated in Fig. 1 in a 9 megabase section of chromosome 9q34 located one megabase from the telomere. The LCN6 gene spans 4.5 kilobases and contains 7 exons (Figs. 1 &2) which is a highly conserved gene structure in the lipocalin family [46]. LCN6 introns generally do not interrupt the coding sequence of beta strands but instead interrupt coding for connecting loops, another conserved feature of lipocalin gene structure [46, 47] (Figs. 2, 3 and 4). Based on the human genome Build 34, Version 1 the gene adjacent to LCN6, 5 kb toward the telomere at LOC138307, Unigene Hs.32991 is similar to the mouse Lcn8 gene [9]. An additional 2.0 kb farther is transcription unit Hs.413902, similar to the rodent Lcn5 gene [26]. Approximately 180 kb toward the telomere from LCN6, is the gene encoding the complement C8 gamma subunit (C8G), and the prostaglandin D2 synthase (PTGDS) gene [48] is located another 30 kb beyond C8G. One megabase (Mb) closer to the centromere than LCN6 are the genes for PAEP [49], odorant binding protein 2A (OBP2A) [50] and LCN1 (tear lipocalin). Another 7 megabases beyond LCN1 closer to the centromere is the LCN2 [51] gene, also known as neutrophil gelatinase-associated lipocalin (NGAL) or in mouse, 24p3. All of these lipocalin genes are expressed in the male reproductive tract [8, 10, 12, 23, 50, 52]. The mouse orthologue of each of these genes is located on mouse chromosome 2.

The open reading frame of human LCN6 encodes a protein of 163 amino acids with a predicted cleavage site releasing a 20 amino acid N-terminal signal peptide [53] (Figs. 2 and 3) and a mature protein with a predicted molecular weight of 16.0 kDa. The three element lipocalin signature motif, GXWY, TDYXXY and R is conserved in rhesus monkey, but R120 is replaced by L120 in human. A ProSite search [54] revealed a consensus cAMP/cGMP-dependent protein kinase phosphorylation site at human and rhesus Ser73, three casein kinase II phosphorylation sites at Ser64, Thr101 and Ser118. No glycosylation sites were predicted. The rhesus LCN6 is 93% identical to the human and includes a 17 amino acid C-terminal extension containing the second cysteine found in many lipocalins, but lacking in the human LCN6 due to the early stop codon position. This stop codon is present in the human genome database (Fig. 2) and was further verified by sequencing several independent RTPCR products derived from different human donors (data not shown). Human LCN6 protein is 40% similar to rat Lcn5 protein [6, 26], 34% to 36% similar to mouse Lcn5 and to human PTGDS [55] and 30% to 32% similar to human LCN2/NGAL [51] and mouse Mup1 [14].

Thus, the similarity of the LCN6 amino acid sequence to other lipocalins is low, nevertheless strong conservation of the lipocalin 3-dimensional structure is predicted by computer analyses. Based on the similarity of the predicted human LCN6 structure to that of mouse Mup1 previously determined by X-ray diffraction [14], a model of the human LCN6 was calculated (Fig. 4). The predicted basket-like β-barrel structure of LCN6 closely matches that of Mup1, however the relatively short C-terminus of human LCN6 lacks the region that in Mup1 contains the cysteine that forms a disulfide bond with a cysteine in the β-strand B. The conserved lipocalin signature amino acids are located close to each other at the closed end of the β-barrels in LCN6 and Mup1. These signature amino acids were proposed to form a receptor-interacting domain [18].

To determine whether LCN6 expression is predominantly in the epididymis as reported for Lcn5 and Lcn8, total RNAs from epididymis and 25 other human tissues and 4 cell lines were analyzed by Northern blot hybridization (Fig. 5). A strongly hybridizing 1.0 kb species was detected only in RNA from epididymis. In addition, an approximately 2.0 kb weakly hybridizing RNA in urinary bladder was near the limits of detection. This larger mRNA in urinary bladder may represent an alternative transcript containing a partial intron or C-terminal extension similar to LCN6 ESTs BI828733 from brain medulla or BG256703 from a testis embryonal carcinoma cell line. These two ESTs are variants that do not correctly encode LCN6. Currently more than 30 ESTs derived from LOC158062 indicate expression in many organs. However many of these fail to encode LCN6 either because they are splicing variants or they originate from the 3', nonLCN6 half of this locus and thus do not indicate LCN6 expression in those tissues.

To determine if LCN6 is regulated by testosterone as reported for Lcn5 in the mouse [56] or testis factors as reported for mouse Lcn8, RNA was obtained from caput, corpus and cauda epididymis of rhesus monkeys that were sham operated, castrated 6 days and castrated but given a single injection of 400 mg testosterone enanthate immediately following testis removal (Fig. 6). The concentration of LCN6 mRNA in caput total RNA samples appeared little affected following 6 days castration and testosterone replacement. Low level mRNA expression in the corpus of the sham operated animal was abolished by castration and not maintained in the testosterone replaced animal suggesting regulation by testis factors other than androgen. This maintenance of similar LCN6 mRNA concentrations under the three treatment conditions contrasts with the androgen regulation of human epididymal (HE) protein 2 [29], epididymis specific clone (ESC) 42 [30] and cystatin (CST) 11 [57] we previously reported in this same experiment. Morning serum testosterone levels prior to testis removal in sham-operated, castrated and castrated-androgen-replaced were 3.0, 3.8 and 1.4 ng/ml respectively. Six days after surgery, morning serum testosterone levels were 2.0, 0.15 and 64.9 ng/ml respectively. Previously reported testosterone levels in rhesus macaques ranged from 5 ng/ml in the morning to 18 ng/ml during the nocturnal surge [58].

LCN6 protein is most abundant in late stage efferent ducts and caput epithelium in human (Fig. 7). LCN6 is also present in the lumen in association with spermatozoa, consistent with secretion from epithelial cells as suggested by the presence of a predicted hydrophobic signal peptide. The protein was also detected at lower levels in corpus and cauda (data not shown). Labeling of human ejaculate spermatozoa with affinity-purified anti-LCN6 demonstrated that LCN6 was present on all spermatozoa (Fig. 8). LCN6 was localized on the head and tail of spermatozoa with the highest concentration of LCN6 seen on the postacrosomal region of the head, where it appeared aggregated into large patches. Along the tail smaller more discrete focal points of LCN6 were observed. Interestingly, the large patches in the postacrosomal region of the head appear in the region of the spermatozoon thought to fuse with the oocyte plasma membrane. The LCN6 protein on spermatozoa is not likely to originate in the testis since the mRNA was not detected by Northern blot analysis in testis RNA (Fig. 5).

LCN6 is the first human member to be described of a gene cluster that includes LCN5 and LCN8 in a lipocalin-rich region of chromosome 9q34. The similar gene clusters on mouse chromosome 2A3 and on rat 3p13 indicate that these genes diverged from a common ancestor prior to the separation of rodent and primate evolutionary lines. The mouse and rat Lcn5 and mouse Lcn8 are the only genes in these conserved clusters previously described. They are exclusively expressed in the epididymis as judged by Northern blot analysis. Our Northern results showing epididymis-specific expression of fully processed LCN6 mRNA might appear to be contradicted by the ESTs derived from LOC158062 that were isolated from libraries from many other organs. This locus represents a fusion of two GenomeScan-predicted models, LCN6 and an adjacent lipocalin-like gene based on BG256703, a bridging EST derived from a testis embryonal carcinoma cell line. However, our results indicate that the LCN6 gene can function independently of the 3' half of the locus, at least in the epididymis. Separate orthologs of LCN6 and of the adjacent gene are predicted in mouse and rat and we isolated the rat Lcn6 as an independently expressed sequence (data not shown). Most of the ESTs listed from this combined locus are derived from the 3' lipocalin-like adjacent gene. The LCN6-specific GenBank ESTs are all from epididymis, "whole testis" which includes epididymis, or different cell lines, thus are consistent with our Northern blot tissue distribution analysis. In addition, GenBank contains LCN6 splicing variant ESTs from brain medulla and prostate. These splicing variants may not be translated, a mechanism previously reported that restricts GnRH expression to the hypothalamus [59].

Expression of LCN6 in distal efferent ducts and caput epididymis and localization of the protein on the surface of spermatozoa are consistent with a role in spermatozoa maturation. That role could be to carry ligands from the proximal epididymis to receptors on distal epithelial cells, a mechanism suggested for Lcn5 transport of retinoic acid [26]. A similar model was proposed to explain the regulation of proenkephalin gene expression in the rat caput by an unidentified spermatozoa-associated factor [60]. Moreover, the ligand might be delivered to receptors in the female tract. Delivery of ligand could result in changes in gene expression in the recipient cells.

The ligand(s) of LCN6 have not been identified, but may be similar to retinoic acid which is a known ligand bound by Lcn5. The Lcn5 and LCN6 proteins have diverged substantially in linear sequence, yet the amino acids known and predicted to line the binding pocket and entrance of LCN6 and rat Lcn5 [6] are more closely related than the 40% similarity of the whole proteins. Of the 23 amino acids that form the ligand binding cavity in rat Lcn5 [61], 9 or 39% are identical in LCN6 and six others are similar for a total of 65% similarity in the ligand binding cavity. Furthermore, 3 of the 5 aromatic amino acids that are deepest in the binding cavity, forming a semicircle surrounding the β-ionone ring of the retinoic acid in the holo-Lcn5, are identical in LCN6. The other two amino acids, Phe6 and Phe11 in Lcn5 are replaced by leucine in LCN6. At the entrance to the binding cavity in Lcn5 are charged amino acids Glu17, Glu63, Arg80, Lys85 and Lys115. These are replaced in LCN6 by Val/Ile, Ser, Ser, Leu and Glu respectively. The effects of these amino acid differences remain to be evaluated by ligand binding studies and X-ray diffraction but presumably the range of ligand structures that can be accommodated in the cavities and their orientations with respect to the amino acids lining the cavity are probably different in LCN6 and Lcn5.

Ligand binding properties of human LCN6 may be affected by the lack of the cysteine near the C-terminus that is strongly conserved in lipocalins. Where present, this cysteine forms a disulfide bridge with a cysteine located in β-strand D and anchors the C-terminus to the β-barrel (Fig 4B). In studies on LCN1, the corresponding intact disulfide bond diminishes the affinity for retinol and restricts the displacement of native lipids by retinol probably by conferring rigidity on the barrel structure and restricting movement of the aromatic side chains involved in ligand binding [62]. Disulfide reduction in LCN1 was accompanied by alteration in ligand-induced conformational changes. The absence of this disulfide bridge in human LCN6 may similarly allow a broader range of ligands to fit into the human pocket than can fit into the monkey form.

The molecular basis of LCN6 interaction with spermatozoa is unknown, but its presence on spermatozoa in discrete patches raises the possibility of interaction with specific receptors. Each known spermatozoon surface lipocalin is distributed according to a particular pattern suggesting a specific molecular interaction. The location of LCN6 generally on post-acrosomal human spermatozoa contrasts with 24p3 on mouse spermatozoa which is predominantly on the anterior acrosomal region [8]. On bull spermatozoa, PTGDS is concentrated on the apical ridge of the acrosomal cap [5]. On surfaces of cell types other than spermatozoa, several lipocalins interact with cell surface receptors [63]. Cellular responses were demonstrated for 24p3/Lcn2 protein which induced apoptosis by a receptor-mediated pathway in murine FL5.12 l2 pro-B lymphocytic cell line [64] and Lcn-1 was internalized after binding a specific receptor on human NT2 embryonal carcinoma cells [65]. Whether internal responses are generated in spermatozoa after lipocalin binding has not been reported.

Regional differences in expression site and differences in the influence of the testis on expression of Lcn5, Lcn8 and LCN6 may be factors driving the evolutionary maintenance in rodent and human of this gene cluster that is derived by gene duplication [9]. The LCN6 expression in distal efferent ducts, initial segment and proximal caput is similar to that of rat Lcn5 [66] and mouse Lcn8 [67] but differs from the mouse Lcn5 which is expressed in distal caput and corpus and accumulates in caudal fluid [4, 26]. The more distal expression of mouse Lcn5 positions it to affect sperm maturation downstream of where Lcn8 becomes available to spermatozoa in the lumen. Unlike the rodent Lcn5, monkey LCN6 mRNA levels appear to be little regulated by androgen and unlike mouse Lcn8, monkey LCN6 regulation by other testis factors, appears minimal 6 days after testis removal. This is a surprising result for an epididymal protein associated with spermatozoa since spermatogenesis and sperm maturation in the epididymis both depend on androgen. The lack of androgen regulation of LCN6 expression may reflect functions apart from sperm maturation such as maintenance of healthy epididymal epithelium under changing physiological conditions. Since a number of lipocalins have demonstrated roles in host defense [68] and the C-terminus of the monkey LCN6 protein is cationic, we tested recombinant human and monkey LCN6 for antibacterial activity and found they had little or no effect on E. coli [33] suggesting that killing bacteria is not a likely role for LCN6.

LCN6 is a novel lipocalin closely related to Lcn5 and Lcn8. The LCN5,6 and 8 genes are likely products of ancient gene duplication events that predate rodent-primate divergence. Predominant expression in the epididymis and location on sperm surface are consistent with a role for LCN6 in male fertility.