A chlamydial type III translocated protein is tyrosine-phosphorylated at the site of entry and associated with recruitment of actin.

The obligate intracellular bacterium Chlamydia trachomatis rapidly induces its own entry into host cells. Initial attachment is mediated by electrostatic interactions to heparan sulfate moieties on the host cell, followed by irreversible binding to an unknown secondary receptor. This secondary binding leads to the recruitment of actin to the site of attachment, formation of an actin-rich, pedestal-like structure, and finally internalization of the bacteria. How chlamydiae induce this process is unknown. We have identified a high-molecular-mass tyrosine-phosphorylated protein that is rapidly phosphorylated on attachment to the host cell. Immunoelectron microscopy studies revealed that this tyrosine-phosphorylated protein is localized to the cytoplasmic face of the plasma membrane at the site of attachment of surface-associated chlamydiae. The phosphoprotein was isolated by immunoprecipitation with the antiphosphotyrosine antibody 4G10 and identified as the chlamydial protein CT456, a hypothetical protein with unknown function. The chlamydial protein (Tarp) appears to be translocated into the host cell by type III secretion because it is exported in a Yersinia heterologous expression assay. Phosphotyrosine signaling across the plasma membrane preceded the recruitment of actin to the site of chlamydial attachment and may represent the initial signal transduced from pathogen to the host cell. These results suggest that C. trachomatis internalization is mediated by a chlamydial type III-secreted effector protein.

The guanine protein coupled receptor rhodopsin is developmentally regulated in the free-living stages of Schistosoma mansoni.

Schistosoma mansoni parasites inhabit three distinct environments including water, intermediate molluscan hosts, and definitive vertebrate hosts. Determining how schistosomes interact with these environments may be one mechanism by which suitable vaccines or novel chemotherapeutic targets will be identified. Towards this end, we describe the identification of a 36-kDa S. mansoni protein that shares extensive sequence similarity to light absorbing rhodopsin guanine protein coupled receptors (GPCRs). This protein, S. mansoni rhodopsin (SmRHO), is the first molecularly characterized GPCR described in schistosomes. Sequence analysis reveals that SmRHO shares extensive phylogenetic conservation among rhodopsins/opsins expressed in water-dwelling invertebrates, possibly indicative of orthology. We demonstrate here that SmRHO is expressed in the free-living, light responsive miracidia and cercaria stages and is down-regulated in the adult, vertebrate residing forms. Moreover, we show that SmRHO is localized to sub-tegumental structures found towards the anterior end of cercariae. As SmRHO may be implicated in schistosome photoreception processes, we have begun a search for additional parasite encoded GPCR super-family members, which may be associated with chemoreception, chemotaxis, and olfaction. Identifying and characterizing new GPCRs may uncover hidden aspects of parasite biology useful towards the development of novel intervention strategies.

Three temporal classes of gene expression during the Chlamydia trachomatis developmental cycle.

The obligate intracellular bacterium Chlamydia trachomatis has a unique developmental cycle that involves functionally and morphologically distinct cell types adapted for extracellular survival and intracellular multiplication. Infection is initiated by an environmentally resistant cell type called an elementary body (EB). Over the first several hours of infection, EBs differentiate into a larger replicative form, termed the reticulate body (RB). Late in the infectious process, RBs asynchronously begin to differentiate back to EBs, which accumulate within the lumen of the inclusion until released from the host cell for subsequent rounds of infection. In an effort to characterize temporal gene expression in relation to the chlamydial developmental cycle, we have used quantitative-competitive polymerase chain reaction (QC-PCR) and reverse transcription (RT)-PCR techniques. These analyses demonstrate that C. trachomatis double their DNA content every 2-3 h, with synthesis beginning between 2 and 4 h after infection. We determined the onset of transcription of specific temporal classes of developmentally expressed genes. RT-PCR analysis was performed on several genes encoding key enzymes or components of essential biochemical pathways and functions. This comparison encompassed approximately 8% of open reading frames on the C. trachomatis genome. In analysis of total RNA samples harvested at 2, 6, 12 and 20 h after infection, using conditions under which a single chlamydial transcript per infected cell is detected, three major temporal classes of gene expression were resolved. Initiation of transcription appears to occur in three temporal classes which we have operationally defined as: early, which are detected by 2 h after infection during the germination of EBs to RBs; mid-cycle, which appear between 6 and 12 h after infection and represent transcripts expressed during the growth and multiplication of RBs; or late, which appear between 12 and 20 h after infection and represent those genes transcribed during the terminal differentiation of RBs to EBs. Collectively, the data suggest that chlamydial early gene functions are weighted toward initiation of macromolecular synthesis and the establishment of their intracellular niche by modification of the inclusion membrane. Surprisingly, representative enzymes of intermediary metabolism and structural proteins do not appear to be transcribed until 10-12 h after infection; coinciding with the onset of observed binary fission of RBs. Late gene functions appear to be predominately those associated with the terminal differentiation of RBs back to EBs.

A tubular network associated with the brush-border surface of the Aedes aegypti midgut: implications for pathogen transmission by mosquitoes.

The mosquito Aedes aegypti is capable of transmitting a variety of pathogens to man and to other vertebrates. The midgut of this insect has been well-studied both as the tissue where the first contact occurs between ingested pathogens and the insect host, and as a model system for blood meal digestion in blood-sucking insects. To understand better the nature of the midgut surface encountered by parasites or viruses, we used scanning electron microscopy to identify the most prominent structures and cell morphologies on the luminal midgut surface. The luminal side of the midgut is a complex and layered set of structures. The microvilli that are found on most, but not all, cells are covered by a network of fine strands that we have termed the microvilli-associated network (MN). The MN strands are membranous, as shown by a membrane bilayer visible in cross sections of MN strands at high magnification in transmission electron micrographs. The MN is found in blood-fed as well as unfed mosquitoes and is not affected by chitinase treatment, suggesting that it is not related to the chitinous peritrophic membrane that is formed only after blood feeding. The cells in the midgut epithelium have two distinct morphologies: the predominant cell type is densely covered with microvilli, while cells with fewer microvilli are found interspersed throughout the midgut. We used lectins to probe for the presence of carbohydrates on the midgut surface. A large number of lectins bind to the luminal midgut surface, suggesting that a variety of sugar linkages are present on the structures visualized by electron microscopy. Some of these lectins partially block attachment of malaria ookinetes to the midgut surface in vitro. Thus, the mosquito midgut epithelium, like the lining of mammalian intestines, is complex, composed of a variety of cell types and extensively covered with surface carbohydrate that may play a role in pathogen attachment.

Persistence of hepatitis C virus in a human megakaryoblastic leukaemia cell line.

Thrombocytopenia is a frequent clinical finding in patients with hepatitis C virus (HCV) infection. Platelets from patients with HCV infection have been identified as carriers of HCV RNA in our previous studies. The present study was designed to further investigate the possibility of HCV replication in megakaryoblasts from which platelets are eventually released. A megakaryoblastic cell line (MEG-01), established from a chronic myelogenous leukaemia patient 13 years ago, was used for this study. The MEG-01 cells were inoculated with fresh serum from a patient with HCV infection and renamed MEG-01-I cells. Surprisingly, both MEG-01 and MEG-01-I were positive by HCV reverse transcription-polymerase chain reaction (RT-PCR) for the existence of HCV RNA and minus-strand HCV RNA, regardless of inoculation. This was further confirmed by in situ RT-PCR. The HCV antigens, such as core, envelope, and non-structural (NS)3 and NS4, were also present in both cell lines, as identified by Western blotting and indirect immunofluorescence staining. In addition, virus-like particles were observed by electron microscopy in the MEG-01 cell line as well as in the MEG-01-I cell line. These findings indicate that the megakaryoblasts are vulnerable to HCV infection and that replication of HCV can occur in these cells. This may help us to better understand the pathogenesis of thrombocytopenia in patients with HCV infection. The MEG-01 cell line, which may have been continuously shedding HCV for years, should be a useful model for experimental research into HCV.

Identification and characterization of a Chlamydia trachomatis early operon encoding four novel inclusion membrane proteins.

Chlamydia trachomatis is a bacterial obligate intracellular parasite that replicates within a vacuole, termed an inclusion, that does not fuse with lysosomes. Within 2 h after internalization, the C. trachomatis inclusion ceases to interact with the endocytic pathway and, instead, becomes fusogenic with exocytic vesicles containing exogenously synthesized NBD-sphingomyelin. Both fusion of exocytic vesicles and long-term avoidance of lysosomal fusion require early chlamydial gene expression. Modification of the chlamydial inclusion probably occurs through the expression and insertion of chlamydial protein(s) into the inclusion membrane. To identify candidate inclusion membrane proteins, antisera were raised against a total membrane fraction purified from C. trachomatis-infected HeLa cells. By indirect immunofluorescence, this antisera recognized the inclusion membrane and, by immunoblot analysis, recognized three chlamydial-specific antigens of approximate molecular weights 15, 18 and 21 kDa. IncG, encoding an 18 kDa and 21 kDa doublet chlamydial antigen, was identified by screening a C. trachomatis, serovar L2, genomic expression library. Three additional genes, incD, incE and incF, were co-transcribed with incG. Monospecific antisera against each of the four genes of this operon demonstrated that the gene products were localized to the chlamydial inclusion membrane. Immediately downstream from the operon containing incD-G was the C. trachomatis homologue of incA. Like IncD, E, F and G, C. trachomatis IncA is also localized to the inclusion membrane. Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis demonstrated that IncD-G, but not incA, are transcribed within the first 2 h after internalization, making them candidates for chlamydial factors required for the modification of the nascent chlamydial inclusion.

The Chlamydia trachomatis IncA protein is required for homotypic vesicle fusion.

Chlamydiae replicate within an intracellular vacuole, termed an inclusion, that is non-fusogenic with vesicles of the endosomal or lysosomal compartments. Instead, the inclusion appears to intersect an exocytic pathway from which chlamydiae intercept sphingomyelin en route from the Golgi apparatus to the plasma membrane. Chlamydial protein synthesis is required to establish this interaction. In an effort to identify those chlamydial proteins controlling vesicle fusion, we have prepared polyclonal antibodies against several Chlamydia trachomatis inclusion membrane proteins. Microinjection of polyclonal antibodies against three C. trachomatis inclusion membrane proteins, IncA, F and G, into the cytosol of cells infected with C. trachomatis demonstrates reactivity with antigens on the cytoplasmic face of the inclusion membrane, without apparent inhibition of chlamydial multiplication. Microinjection of antibodies against the C. trachomatis IncA protein, however, results in the development of an aberrant multilobed inclusion structure remarkably similar to that of C. psittaci GPIC. These results suggest that the C. trachomatis IncA protein is involved in homotypic vesicle fusion and/or septation of the inclusion membrane that is believed to accompany bacterial cell division in C. psittaci. This proposal is corroborated by the expression of C. trachomatis and C. psittaci IncA in a yeast two-hybrid system to demonstrate C. trachomatis, but not C. psittaci, IncA interactions. Despite the inhibition of homotypic fusion of C. trachomatis inclusions, fusion of sphingomyelin-containing vesicles with the inclusion was not suppressed.

Evaluation of the role of the Yersinia pestis plasminogen activator and other plasmid-encoded factors in temperature-dependent blockage of the flea.

Yersinia pestis, the plague bacillus, has a plasminogen activator (pla) gene on the 9.5-kb plasmid pPla that is hypothesized to play a role in producing the foregut blockage in the flea vector that precedes transmission. In this study, however, Y. pestis that lacked pPla, the 70-kb virulence plasmid, or both plasmids, proved able to block Xenopsylla cheopis fleas normally. Blockage rates decreased with increasing environmental temperature for fleas infected with either wild type or pPla- Y. pestis. Thus, procoagulant ability of the Y. pestis pla gene product does not mediate blockage, nor does its ability to induce fibrinolysis at>28 degreesC account for failure to block at elevated temperatures. A Y. pestis strain that lacked all or part of the third plasmid of 110 kb, however, failed to colonize the flea midgut normally, indicating that one or more genes on the large plasmid may be required for vectorborne transmission.

Adhesion of Plasmodium gallinaceum ookinetes to the Aedes aegypti midgut: sites of parasite attachment and morphological changes in the ookinete.

Plasmodium gallinaceum ookinetes adhered to Aedes aegypti midgut epithelia when purified ookinetes and isolated midguts were combined in vitro. Ookinetes preferentially bound to the microvillated luminal surface of the midgut, and they seemed to interact with three types of structures on the midgut surface. First, they adhered to and migrated through a network-like matrix, which we have termed microvilli-associated network, that covers the surface of the microvilli. This network forms on the luminal midgut surface in response to blood or protein meals. Second, the ookinetes bound directly to the microvilli on the surface of the midgut and were occasionally found immersed in the thick microvillar layer. Third, the ookinetes associated with accumulations of vesicular structures found interspersed between the microvillated cells of the midgut. The origin of these vesicular structures is unknown, but they correlated with the surface of midgut cells invaded by ookinetes as observed by TEM. After binding to the midgut, ookinetes underwent extensive morphological changes: they frequently developed one or more annular constrictions, and their surface roughened considerably, suggesting that midgut components remain bound to the parasite surface. Our observations suggest that, in a natural infection, the ookinete interacts in a sequential manner with specific components of the midgut surface. Initial binding to the midgut surface may activate the ookinete and cause morphological changes in preparation for invasion of the midgut cells.

Expression of GTPase-deficient Gialpha2 results in translocation of cytoplasmic RGS4 to the plasma membrane.

The members of a recently identified protein family termed regulators of G-protein signaling (RGS) act as GTPase-activating proteins for certain Galpha subunits in vitro, but their physiological effects in cells are uncertain in the face of similar biochemical activity and overlapping patterns of tissue expression. Consistent with its activity in in vitro GTPase-activating protein assays, RGS4 interacts efficiently with endogenous proteins of the Gi and Gq subclasses of Galpha subunits but not with G12alpha or Gsalpha. Unlike other RGS proteins such as RGS9, RGS-GAIP, and Sst2p, which have been reported to be largely membrane-associated, a majority of cellular RGS4 is found as a soluble protein in the cytoplasm. However, the expression of a GTPase-deficient Gialpha subunit (Gialpha2-Q204L) resulted in the translocation of both wild type RGS4 and a non-Gialpha-binding mutant (L159F) to the plasma membrane. These data suggest that RGS4 may be recruited to the plasma membrane indirectly by G-protein activation and that multiple RGS proteins within a given cell might be differentially localized to determine a physiologic response to a G-protein-linked stimulus.

Characterization of an endosymbiont infecting wood ticks, Dermacentor andersoni, as a member of the genus Francisella.

A microorganism (Dermacantor andersoni symbiont [DAS]) infecting Rocky Mountain wood ticks (D. andersoni) collected in the Bitterroot Mountains of western Montana was characterized as an endosymbiont belonging to the genus Francisella. Previously described as Wolbachia like, the organism's DNA was amplified from both naturally infected tick ovarial tissues and Vero cell cultures by PCR assay with primer sets derived from eubacterial 16S ribosomal DNA (rDNA) and Francisella membrane protein genes. The 16S rDNA gene sequence of the DAS was most similar (95.4%) to that of Francisella tularensis subsp. tularensis. Through a combination of Giménez staining, PCR assay, and restriction fragment length polymorphism analysis, 102 of 108 female ticks collected from 1992 to 1996 were infected. Transovarial transmission to female progeny was 95.6%, but we found no evidence of horizontal transmission.

Rickettsia rickettsii growth and temperature-inducible protein expression in embryonic tick cell lines.

Rickettsia rickettsii has limited adverse effects on its arthropod vector, but causes severe disease in man. To model differences in host-parasite interaction, R. rickettsii growth and protein expression were examined at temperatures reflective of host environment in the tick cell lines DALBE3 and IDE2, the human endothelial cell line ECV304, and the African green monkey kidney cell line Vero76. At low multiplicities of infection, rickettsial titres increased 10(2)-10(3)-fold in all cell lines after incubation for 3 days at 34 degrees C. At higher multiplicities and with extended incubation, R. rickettsii showed enhanced survival in tick versus mammalian cells. No difference in rickettsial ultrastructure or protein profiles was detected between different host cell types. Rickettsial proteins of 42, 43, 48, 75 and 100 kDa are induced in tick cells shifted from 28 degrees to 34 degrees C, but not in cells maintained at 28 degrees C. This temperature response may be associated with expression of rickettsial determinants that are pathogenic to mammalian hosts.

Invasion and cytopathic killing of human lymphocytes by spirochetes causing Lyme disease.

Lyme disease is a persistent low-density spirochetosis caused by Borrelia burgdorferi sensu lato. Although spirochetes causing Lyme disease are highly immunogenic in experimental models, the onset of specific antibody responses to infection is often delayed or undetectable in some patients. The properties and mechanisms mediating such immune avoidance remain obscure. To examine the nature and consequences of interactions between Lyme disease spirochetes and immune effector cells, we coincubated B. burgdorferi with primary and cultured human leukocytes. We found that B. burgdorferi actively attaches to, invades, and kills human B and T lymphocytes. Significant killing began within 1 hour of mixing. Cytopathic effects varied with respect to host cell lineage and the species, viability, and degree of attenuation of the spirochetes. Both spirochetal virulence and lymphocytic susceptibility could be phenotypically selected, thus indicating that both bacterial and host cell factors contribute to such interactions. These results suggest that invasion and lysis of lymphocytes may constitute previously unrecognized factors in Lyme disease and bacterial pathogenesis.

Origins and functions of the chlamydial inclusion.

Chlamydiae dissociate themselves from the endocytic pathway shortly after internalization by actively modifying the vacuole to become fusogenic with sphingomyelin-containing exocytic vesicles. Interaction with this secretory pathway appears to provide a pathogenic mechanism that allows chlamydiae to establish themselves in a site that is not destined to fuse with lysosomes.

Rickettsia peacockii sp. nov., a new species infecting wood ticks, Dermacentor andersoni, in western Montana.

Rickettsia peacockii, a new species of spotted fever group rickettsiae, was identified from Rocky Mountain wood ticks (Dermacentor andersoni) collected in the Sapphire Mountain Range on the eastern side of Bitterroot Valley, Montana. DNA from R. peacockii SkalkahoT (T = type strain) in naturally infected tick tissue was amplified by a PCR assay with primer sets derived from eubacterial 16S ribosomal DNA (rDNA), rickettsial citrate synthase, and 190-kDa surface antigen (rOmpA) genes. Partial 16S rDNA and rOmpA gene sequences exhibited levels of similarity of 99.7 and 93.2%, respectively, with the sequences of the spotted fever agent Rickettsia rickettsii R. By using Gimenez staining, fluorescent antibody tests, a PCR assay, and a restriction fragment length polymorphism analysis, 76 of 115 female ticks (minimal field infection rate, 66.1%) collected between 1992 and 1995 were found to be infected. The organism is passed transstadially and transovarially (minimal vertical transmission rate, 73.3%), and infections are localized in ovarial tissues. Attempts to cultivate R. peacockii were unsuccessful.

Vesicular interactions of the Chlamydia trachomatis inclusion are determined by chlamydial early protein synthesis rather than route of entry.

Chlamydiae replicate intracellularly within a vacuole that has recently been characterized as intersecting an exocytic pathway. One of the initial events during chlamydial infection is the expression of a chlamydial early gene product(s) that effectively isolates the inclusion from the endocytic-lysosomal pathway and makes it fusogenic with sphingomyelin-containing exocytic vesicles. Associated with this change in vesicular interaction is the delivery of the vacuole to the peri-Golgi region of the host cell. Inhibition of chlamydial early transcription or translation causes Chlamydia trachomatis-containing vesicles to remain dispersed throughout the cytoplasm, where they eventually fuse with lysosomes. Chlamydiae that have been internalized by Fc-mediated endocytosis also avoid lysosomal digestion by a mechanism that requires chlamydial protein synthesis. These results suggest that the vesicular interactions of the chlamydial inclusion are defined by parasite-directed modification of the endocytic vesicle rather than by the route of internalization.

Temporal analysis of the developing Chlamydia psittaci inclusion by use of fluorescence and electron microscopy.

The chlamydiae are obligate intracellular parasites that develop and multiply within a vacuole (termed an inclusion) that does not fuse with lysosomes. Inclusion morphology varies dramatically among the different chlamydiae, particularly within the species Chlamydia psittaci. Some strains develop within a single vacuole, while the mature inclusion of other strains consists of several distinct lobes, each filled with chlamydial developmental forms. The development of this lobed structure was investigated in HeLa cells infected with the guinea pig inclusion conjunctivitis (GPIC) strain of C. psittaci. We employed two recently described probes for the chlamydial inclusion to study the development of these unique lobed structures. The novel probes were an antiserum directed at a protein localized to the GPIC inclusion membrane (anti-IncA) and the fluorescent sphingolipid (N-[7-(4-nitrobenzo-2-oxa-1,3-)]) aminocaproyl sphingosine (NBD-ceramide). Lobed inclusions developed in cells infected at very low multiplicities of infection, suggesting that the structure is not a function of infection by more than one elementary body (EB). Double-label fluorescent-antibody analysis with anti-IncA and an antibody directed at a chlamydial outer membrane protein showed that, prior to 18 h postinfection (p.i.), the inclusion membrane and the chlamydial membrane were tightly associated. After 18 to 20 h p.i., the lobes began to expand and fill with developmental forms and the inclusion membrane and chlamydial membrane became distinct. At times from 8 to 48 h p.i., GPIC inclusions were shown to receive fluorescent derivatives of NBD-ceramide and to be localized to the perinuclear region of the host cell. Labeled lectins with affinity for carbohydrate moieties localized to the Golgi apparatus showed that the lobes of mature inclusions surround the Golgi apparatus. Labeling with NBD-ceramide and the Golgi apparatus-specific lectins therefore demonstrated a functional and physical association of the inclusion with the Golgi apparatus throughout the developmental cycle. Collectively, these results lead to a model for the development of the lobed chlamydial inclusion. We propose that the lobed structure is a result of division of inclusions occurring in parallel with the multiplication of reticulate bodies (RB) early in the developmental cycle. The division of inclusions slows or stops in mid-cycle, and dividing RB accumulate within the enlarging lobes. The RB then differentiate to EBs, the inclusion and cell are lysed, and EBs are freed to infect another cell.

Sphingolipids and glycoproteins are differentially trafficked to the Chlamydia trachomatis inclusion.

Chlamydia trachomatis is an obligate intracellular pathogen that multiples within the confines of a membrane-bound vacuole called an inclusion. Approximately 40-50% of the sphingomyelin synthesized from exogenously added NBD-ceramide is specifically transported from the Golgi apparatus to the chlamydial inclusion (Hackstadt, T., M.A. Scidmore, and D.D. Rockey. 1995. Proc. Natl. Acad. Sci. USA. 92: 4877-4881). Given this major disruption of a cellular exocytic pathway and the similarities between glycolipid and glycoprotein exocytosis, we wished to determine whether the processing and trafficking of glycoproteins through the Golgi apparatus to the plasma membrane in chlamydia-infected cells was also disrupted. We analyzed the processing of several model glycoproteins including vesicular stomatitis virus G-protein, transferrin receptor, and human histocompatibility leukocyte class I antigen. In infected cells, the posttranslational processing and trafficking of these specific proteins through the Golgi apparatus and subsequent transport to the plasma membrane was not significantly impaired, nor were these glycoproteins found associated with the chlamydial inclusion membrane. Studies of receptor recycling from endocytic vesicles employing fluorescently and HRP-tagged transferrin and anti-transferrin receptor antibody revealed an increased local concentration of transferrin and transferrin receptor around but never within the chlamydial inclusion. However, Scatchard analysis failed to show either an increased intracellular accumulation of transferrin receptor or a decreased number of plasma membrane receptors in infected cells. Furthermore, the rate of exocytosis from the recycling endosomes to the plasma membrane was not altered in chlamydia-infected cells. Thus, although C. trachomatis disrupts the exocytosis of sphingolipids and the Golgi apparatus appears physically distorted, glycosylation and exocytosis of representative secreted and endocytosed proteins are not disrupted. These results suggest the existence of a previously unrecognized sorting of sphingolipids and glycoproteins in C. trachomatis-infected cells.

A rapid assay for detecting cellular TdT enzymatic activity.

We have developed a solid-phase assay for the quantification of terminal deoxynucleotidyl transferase (TdT) enzymatic activity in crude cellular extracts. Affinity-purified, polyclonal anti-TdT antibodies are bound to the wells of a microtiter plate, and TdT in extracts is then bound to the immobilized antibodies. The enzymatic activity of the antibody-bound TdT is measured directly in the wells of the microtiter plate. This method yields highly reproducible values, even with samples of low activity. Because it is also technically very simple, it is ideal for determining enzymatic activity for large numbers of clones with limited numbers of cells.

Interpatient variation in response to subcutaneous versus intravenous low dose erythropoietin.

Subcutaneous (SC) recombinant human erythropoietin (EPO) has been reported to correct anemia in hemodialysis patients at lower doses than intravenous (IV) EPO. Those trials involved relatively high doses of EPO or did not control adequately for time-related falls in dose requirements. Therefore, on open-label double-crossover study was performed to compare the hemoglobin (Hb) response to low dose SC versus IV EPO. Ten (4 male) maintenance hemodialysis patients previously stabilised on low dose EPO for 18 +/- 3 months (mean +/- SEM) were given EPO IV for 12 weeks (IV#1), then SC for 24 weeks and then IV for a further 20 weeks (IV#2). Iron status and other factors known to modify response to EPO were kept constant. EPO dose was not changed unless Hb rose above 100 g/l, when the dose was reduced to keep Hb between 90 and 100 g/l. Initial EPO dose was 64 +/- 10 u/kg/week. Mean Hb, measured monthly, was not different during the 3 treatment periods. There was wide interpatient variation in the relative response to IV versus SC EPO. Mean Hb was higher on IV EPO in 5 patients (by 6.1 +/- 2.0 g/l) and higher on SC EPO in 5 patients (by 12.1 +/- 4.1 g/l). The difference in mean Hb during IV versus SC administration was more than 5 g/l in 6 patients, being higher in 3 patients during IV administration (by 8.7 +/- 4.6 g/l) and in 3 during SC (by 17.4 +/- 4.6 g/l). In conclusion, the more efficient route of administration of EPO is not predictable for individual patients, and should be sought to allow possible dose reduction.