72
if platelet consumption was thought to be increased, including fever of over 38.5~ accompanied by a rapid decrease in platelet count, coagulation factor deficiencies due to sepsis or leukemia, or the presence of hyperleukocytosis. All centers used the same chemotherapeutic regimens, as well as standardized prophylactic antimicrobials and management of infections. Hemoglobin was maintained at over 80 g/L. Patients had daily platelet counts and received either pooled random donor ABO-compatible platelets or apheresis platelets determined by the standard practice at each center and by availability. The study end points were bleeding complications, classified according to World Health Organization (WHO) severity criteria, and the number of platelet and red cell transfusions administered. One hundred five consecutive patients undergoing 216 cycles of chemotherapy with 3,843 days of thrombocytopenia (platelet count less than 25 • 109/L) were enrolled. There was some imbalance in the groups, because platelet counts before each cycle of chemotherapy were significantly lower in the group A patients, with a longer mean duration of thrombocytopenia. Bleeding complications occurred in 18.2% of chemotherapy cycles in group A and in 17.0% of cycles in group B, a nonsignificant difference. However, severe bleeding (WHO grades 3 and 4) occurred only in seven patients in group B and was related to sepsis with DIC rather than to platelet count. Two of these patients died of septic shock. Eighty-six percent of the severe bleeding episodes occurred during induction chemotherapy cycles. The proportion of thrombapheresis to random donor platelet concentrates transfused was the same in both groups (1:5), as was the number of red cell transfusions. However, the number of platelet transfusions was decreased by one third in group A (mean of three apheresis and 15.4 random donor platelet concentrates per treatment cycle) compared with group B (mean of 4.8 apheresis and 25.4 random donor platelet concentrates per treatment cycle). The strengths of this study include the relatively large number of patients and days of thrombocytopenia at several different centers. Weaknesses include the study design, which is not randomized or blinded and therefore led to imbalances between the study groups and potential bias. It also would be interesting to know the time delay between blood being drawn for performance of the platelet count and platelets arriving at the patient's bedside, because this delay may be of concern to centers contemplating changes to their platelet transfusion policy. Study conclusions are similar to that of Rebulla et al (N Engl J Med 337:1870-1875, 1997) supporting the safety of a lower transfusion trigger for AML patients. (M. G.)
Risk factors for hepatocellular carcinoma and its incidence after interferon treatment in patients with chronic hepatitis C. Kasahara A, Hayashi N, Mochizuki K, et aL Hepatology 27:1394-1402, 1998. Determinants of outcome of compensated hepatitis C virus related cirrhosis. Serfaty L, Aumaffre H, Chazouillores O, et al. Hepatology 27:1435-1440, 1998. Few studies have observed patients for a sufficiently long period to determine the influence of interferon treatment on the natural history of hepatitis C virus (HCV) infection, including the development of symptomatic cirrhosis and hepatocellular carcinoma (HCC). These studies suggest a long-term benefit of interferon treatment in patients without cirrhosis, with a bio-
CURRENT LITERATURE chemical response to treatment (Kasahara et al), and in patients with compensated cirrhosis, regardless of biochemical response (Serfaty et al). Kasahara et al report the results of a prospective study of 1,022 HCV patients with a low initial prevalence of cirrhosis (3%), treated with various interferon protocols in Japan. The mean follow-up period was 36 months. Three hundred thirteen patients had a sustained response to interferon (alanine transaminase [ALT] levels remained normal for more than 24 weeks after therapy), 304 had a transient response, and 405 had no enzyme response. HCC developed in 46 patients, with a 7.9-fold higher risk for nonresponders compared with sustained responders and an intermediate risk for unsustained responders. The risk of development of cirrhosis also varied with response to treatment, with a 6.18-fold higher risk in nonresponders compared with responders. The second study, by Serfaty et al in France, followed the incidence of liver decompensation (defined as the development of ascites, jaundice, variceal bleeding, or encephalopathy) and of HCC in a population of 103 HCV-positive patients with compensated cirrhosis. The mean follow-up period was 40 months. Fifty-nine patients were treated with interferon, and 44 were not; most patients seen before 1992 were started on interferon, whereas those seen later were not, because of the low rate of biochemical response noted in earlier studies. Baseline characteristics were similar in both groups. In multivariate analysis, absence of interferon treatment was the only independent factor predictive of the development of HCC or decompensation. At 4 years, the rate of HCC was 4.4% in treated versus 23% in untreated patients. Survival at 2 and 4 years was significantly higher in the treated than in the untreated patients (97% and 92%, v 95% and 63%, respectively). The impact of interferon was independent of the biochemical response, and only 10% of treated patients had shown a sustained response to interferon treatment. The authors hypothesized that interferon may have an antifibrogenic effect on the liver independent of its antiviral properties. An accompanying editorial by Poynard and Opolon (pp 1443 to 1444) points out a major weakness of both studies, the lack of a randomized control group, but suggests that use of a control group now may be unethical given the increasing data on the utility of interferon treatment. (M. G.)
Association of XK and Kell blood group proteins. Russo D, Redman C, Lee S. J Biol Chem 273:1395013956, 1998. An association of the Kell glycoprotein and the XK protein in red blood cell membranes was predicted from early serological studies in which a phenotypic relationship was noticed between antigens of the Kell blood group system and the Kx antigen. In the K0 phenotype, in which there are no detectable Kell surface antigens, there is an enhanced expression of Kx antigens. In contrast, in the McLeod phenotype, there is an absence of Kx, an otherwise ubiquitous red cell antigen, and a depressed expression of antigens in the Kell blood group system. Also, treatment of normal red blood cells with reducing agents inactivated Kell antigens and enhanced Kx antigens, indicating a sulfhydryl involvement in the presentation of Kell and Kx antigens on red blood cells. Moreover, although K0 red cells have a higher Kx activity than normal, they contain less XK protein, suggesting that a lack of Kell protein may expose more Kx antigens on the
CURRENT LITERATURE surface of the red cell. Kell is a type II membrane glycoprotein that carries over 20 blood group antigens and is classified as a member of the neprilysin family of zinc metallopeptidases. XK is a type III (multipass) membrane protein that carries the Kx antigen and has structural features that suggest it is a transport protein. Kell has 16 (15 extracellular) cysteine residues, and XK has 11 (1 extracellular) cysteine residues. In this current report, Russo et al show that the Kell glycoprotein and the XK protein are associated in the red cell membrane by a disulfhydyl bond. Co-expression of a combination of wild-type and mutated proteins (by site-directed mutagenesis of cDNA) in transfected COS-1 cells showed that Cys 72 of Kell is cross-linked to Cys 347 of XK. Because Cys 72 of the Kell protein is known to be extracellular (Russo et al, Blood 1994;84:3518-3523), its interaction with Cys 347 of XK protein indicates that the disulfide linkage of Kell to XK lies close to the extracellular membrane surface of the red blood cell and supports the predicted topology of XK. Furthermore, the authors show by chemical cross-linking of red cell membranes with dithiobispropionimidate that glyceraldehyde-3-phosphate dehydrogenase is a near neighbor of Kell. This article describes in detail the experiments used to show the nature of the association of a multipass and a single pass protein in the red cell membrane. Such information adds to our understanding of the architecture of membranes. (M.E.R.)
Complete deficiency of glycophorin A in red blood cells from mice with targeted inactivation of the band 3 (AE1) gene. Hassoun H, Hanada 7;, Lutchman M, et al. Blood 91:2146-2151, 1998. This paper provides information about the importance of a multipass membrane protein to the presence of a single-pass protein in the red cell membrane. A large body of evidence supports the notion that band 3 interacts with glycophorin A (GPA) in the red cell membrane. In human GPA-deficient red blood cells (RBCs), band 3 is more extensively glycosylated. Access to band 3-deficient mice allowed the authors of this paper to show that GPA (and protein 4.2) is also absent from the RBCs. The mice were generated by selective inactivation of the AE1 anion exchanger gene. Membranes from RBCs of the band 3 deficient mice possess apparently normal amounts of protein 4.1, adducin, dematin, p55, and glycophorin C, but were devoid of protein 4.2 and GPA. This implies that in band 3, protein 4.2 and GPA may form a tertiary complex in the RBC membrane. GPA was shown to be absent from the band 3-deficient RBCs by Western blot and immunocytochemistry techniques using a polyclonal anti-GPA (human) that was shown to cross-react with the GPA homolog in mouse RBCs. A polymerase chain reaction assay was used to confirm the presence of GPA mRNA, and pulse-label and pulse-chase experiments showed that GPA was not incorporated in the RBC membrane and that it was rapidly degraded in the cytoplasm. Based on their findings and other published evidence, the authors propose that band 3 plays a chaperone-like role, which is necessary for the recruitment of GPA to the RBC membrane. The loss of GPA and protein 4.2 in band 3-deficient RBCs raises interesting questions about the / role of band 3 and protein 4.2 on the kinetics of transport, assembly, expression, and processing of GPA in the RBC membrane. (M.E.R. )
73 Paranodin, a glycoprotein of neuronal paranodal membranes. Menegoz M, Gaspar P, Le Bert M, et al. Neuron 19:319-331, 1997. This article describes the purification of a glycoprotein called Paranodin and the cloning of the gene that encodes the protein. Paranodin is a 180-Kd transmembrane neuronal glycoprotein that is concentrated in axonaI membranes at their junctions with myelinating glial cell, and in paranodes of nerve fibers (both central and peripheral). Paranodin plays a critical role in the physiological properties of myelinated nerve fibers. The large extracellular domain of paranodin is related to neurexins, and its short intracellular tail binds protein 4.1, a membrane skeletonanchoring protein. Paranodin may be a critical component of the macromolecular complex involved in the tight interactions between axons and myelinating glial cells characteristic of the paranodal region. Interestingly, the putative intracellular region of paranoidin (residues 1308 to 1381) contains a short transmembrane segment, which shows striking homology (10 residues in a stretch of 14 are identical) with the sequence located at a similar position in glycophorin C. This region of glycophorin C (glycophorin C carries antigens in the Gerbich blood group system) has been shown to bind to human red blood cell (RBC) protein 4.1 in the junctions of spectrin to actin. Thus, paranodin provides another example of a protein in RBCs that has a homolog in brain. (M.E.R.)
Characterization of the gene encoding the human Kidd blood group/urea transporter protein. Lucien IV, Sidoux-Walter F, Olives B, et al. J Biol Chem 273:1297312980, 1998. Antigens in the Kidd blood group are carried on an integral membrane glycoprotein that is involved in the transport of urea through the red cell membrane. This glycoprotein is also present on endothelial cells of the vasa recta in the inner and outer medulla of the kidney. This article characterizes the exon-intron structure of human blood group Kidd (urea transporter) gene. The gene consists of 11 exons distributed over 30 kilobases. The transcription initiation site is located 335 base pairs upstream of the translation start point located in exon 4. The 5'-flanking region, from nucleotide - 837 to - 336, contains TATA, inverted CAAT boxes, and GATA-1/SP1 erythroid-specific cis-acting regulatory elements. Analysis of the 3'-untranslated region indicates that the two equally abundant erythroid transcripts of 4.4 and 2.0 kilobase pairs arise from the use of different alternative polyadenylation signals. Blood samples from two individuals (B.S. and L.E) with J k ( a - b - ) red blood cells (RBCs) were obtained and analyzed. In both individuals, the Kidd gene was grossly normal. Further analysis, including sequencing showed single-point mutations in the intronic splice motifs. In B.S., the invariant " g " residue of the 3'-accepter splice site of intron was changed to "a," causing altered splicing events. Transcripts lacking either exon 6, exons 6 and 8, or exons 6, 7, 8, and 9 were observed. In L.R, the invariant " g " residue of the 5'-donor splice site was mutated to "t," resulting in transcripts lacking either exon 7 or exons 7, 8, and 9. The protein isoforms potentially encoded by the spliced transcripts were not detected by Western blot analysis. Expression studies in Xenopus oocytes showed that the truncated proteins encoded by the spliced transcripts were not expressed on the oocytes's