Transferability of neural network clinical deidentification systems

OBJECTIVE

Neural network deidentification studies have focused on individual datasets. These studies assume the availability of a sufficient amount of human-annotated data to train models that can generalize to corresponding test data. In real-world situations, however, researchers often have limited or no in-house training data. Existing systems and external data can help jump-start deidentification on in-house data; however, the most efficient way of utilizing existing systems and external data is unclear. This article investigates the transferability of a state-of-the-art neural clinical deidentification system, NeuroNER, across a variety of datasets, when it is modified architecturally for domain generalization and when it is trained strategically for domain transfer.

MATERIALS AND METHODS

We conducted a comparative study of the transferability of NeuroNER using 4 clinical note corpora with multiple note types from 2 institutions. We modified NeuroNER architecturally to integrate 2 types of domain generalization approaches. We evaluated each architecture using 3 training strategies. We measured transferability from external sources; transferability across note types; the contribution of external source data when in-domain training data are available; and transferability across institutions.

RESULTS AND CONCLUSIONS

Transferability from a single external source gave inconsistent results. Using additional external sources consistently yielded an F1-score of approximately 80%. Fine-tuning emerged as a dominant transfer strategy, with or without domain generalization. We also found that external sources were useful even in cases where in-domain training data were available. Transferability across institutions differed by note type and annotation label but resulted in improved performance.

MT-clinical BERT: scaling clinical information extraction with multitask learning

OBJECTIVE

Clinical notes contain an abundance of important, but not-readily accessible, information about patients. Systems that automatically extract this information rely on large amounts of training data of which there exists limited resources to create. Furthermore, they are developed disjointly, meaning that no information can be shared among task-specific systems. This bottleneck unnecessarily complicates practical application, reduces the performance capabilities of each individual solution, and associates the engineering debt of managing multiple information extraction systems.

MATERIALS AND METHODS

We address these challenges by developing Multitask-Clinical BERT: a single deep learning model that simultaneously performs 8 clinical tasks spanning entity extraction, personal health information identification, language entailment, and similarity by sharing representations among tasks.

RESULTS

We compare the performance of our multitasking information extraction system to state-of-the-art BERT sequential fine-tuning baselines. We observe a slight but consistent performance degradation in MT-Clinical BERT relative to sequential fine-tuning.

DISCUSSION

These results intuitively suggest that learning a general clinical text representation capable of supporting multiple tasks has the downside of losing the ability to exploit dataset or clinical note-specific properties when compared to a single, task-specific model.

CONCLUSIONS

We find our single system performs competitively with all state-the-art task-specific systems while also benefiting from massive computational benefits at inference.

Whole-exome sequencing is a valuable diagnostic tool for inherited peripheral neuropathies: Outcomes from a cohort of 50 families

The inherited peripheral neuropathies (IPNs) are characterized by marked clinical and genetic heterogeneity and include relatively frequent presentations such as Charcot-Marie-Tooth disease and hereditary motor neuropathy, as well as more rare conditions where peripheral neuropathy is associated with additional features. There are over 250 genes known to cause IPN-related disorders but it is estimated that in approximately 50% of affected individuals a molecular diagnosis is not achieved. In this study, we examine the diagnostic utility of whole-exome sequencing (WES) in a cohort of 50 families with 1 or more affected individuals with a molecularly undiagnosed IPN with or without additional features. Pathogenic or likely pathogenic variants in genes known to cause IPN were identified in 24% (12/50) of the families. A further 22% (11/50) of families carried sequence variants in IPN genes in which the significance remains unclear. An additional 12% (6/50) of families had variants in novel IPN candidate genes, 3 of which have been published thus far as novel discoveries (KIF1A, TBCK, and MCM3AP). This study highlights the use of WES in the molecular diagnostic approach of highly heterogeneous disorders, such as IPNs, places it in context of other published neuropathy cohorts, while further highlighting associated benefits for discovery.

Challenges and practical approaches with word sense disambiguation of acronyms and abbreviations in the clinical domain

Objectives

Although acronyms and abbreviations in clinical text are used widely on a daily basis, relatively little research has focused upon word sense disambiguation (WSD) of acronyms and abbreviations in the healthcare domain. Since clinical notes have distinctive characteristics, it is unclear whether techniques effective for acronym and abbreviation WSD from biomedical literature are sufficient.

Methods

The authors discuss feature selection for automated techniques and challenges with WSD of acronyms and abbreviations in the clinical domain.

Results

There are significant challenges associated with the informal nature of clinical text, such as typographical errors and incomplete sentences; difficulty with insufficient clinical resources, such as clinical sense inventories; and obstacles with privacy and security for conducting research with clinical text. Although we anticipated that using sophisticated techniques, such as biomedical terminologies, semantic types, part-of-speech, and language modeling, would be needed for feature selection with automated machine learning approaches, we found instead that simple techniques, such as bag-of-words, were quite effective in many cases. Factors, such as majority sense prevalence and the degree of separateness between sense meanings, were also important considerations.

Conclusions

The first lesson is that a comprehensive understanding of the unique characteristics of clinical text is important for automatic acronym and abbreviation WSD. The second lesson learned is that investigators may find that using simple approaches is an effective starting point for these tasks. Finally, similar to other WSD tasks, an understanding of baseline majority sense rates and separateness between senses is important. Further studies and practical solutions are needed to better address these issues.

Evaluating measures of semantic similarity and relatedness to disambiguate terms in biomedical text

INTRODUCTION

In this article, we evaluate a knowledge-based word sense disambiguation method that determines the intended concept associated with an ambiguous word in biomedical text using semantic similarity and relatedness measures. These measures quantify the degree of similarity or relatedness between concepts in the Unified Medical Language System (UMLS). The objective of this work is to develop a method that can disambiguate terms in biomedical text by exploiting similarity and relatedness information extracted from biomedical resources and to evaluate the efficacy of these measure on WSD.

METHOD

We evaluate our method on a biomedical dataset (MSH-WSD) that contains 203 ambiguous terms and acronyms.

RESULTS

We show that information content-based measures derived from either a corpus or taxonomy obtain a higher disambiguation accuracy than path-based measures or relatedness measures on the MSH-WSD dataset.

AVAILABILITY

The WSD system is open source and freely available from http://search.cpan.org/dist/UMLS-SenseRelate/. The MSH-WSD dataset is available from the National Library of Medicine http://wsd.nlm.nih.gov.

Geographical Distribution and Survival of Iris yellow spot virus in Spiny Sowthistle, Sonchus asper, in Georgia

Iris yellow spot virus (IYSV) has occurred in Georgia since 2003. IYSV is transmitted by onion thrips, Thrips tabaci. During a weed survey in the Vidalia onion-growing zone (VOZ), spiny sowthistle (Sonchus asper) was identified as a host for IYSV. Spiny sowthistle is widespread in Georgia, and this presented an opportunity to study the natural spread of IYSV and assess its potential role in IYSV epidemiology. From 2007 to 2009, during the spring season, 2,011 sowthistle samples were collected from various counties within and outside the VOZ. The samples were tested for IYSV infection by enzyme-linked immunosorbent assay and confirmed by reverse-transcription polymerase chain reaction and sequencing. IYSV sequences from sowthistle were 98 to 99% identical to onion IYSV sequences from onion originated from Georgia. By the third year, IYSV-infected sowthistle plants were found in 79% of the counties in the VOZ and in 61% of the sampled counties in all directions, except to the east of the VOZ. Furthermore, thrips-mediated transmission assays confirmed that T. tabaci can efficiently transmit IYSV from onion to sowthistle. Sowthistle also supported T. tabaci survival and reproduction. These findings demonstrate that sowthistle plants can serve as an IYSV inoculum source and as a thrips reservoir.

Semantic Similarity and Relatedness between Clinical Terms: An Experimental Study

Automated approaches to measuring semantic similarity and relatedness can provide necessary semantic context information for information retrieval applications and a number of fundamental natural language processing tasks including word sense disambiguation. Challenges for the development of these approaches include the limited availability of validated reference standards and the need for better understanding of the notions of semantic relatedness and similarity in medical vocabulary. We present results of a study in which eight medical residents were asked to judge 724 pairs of medical terms for semantic similarity and relatedness. The results of the study confirm the existence of a measurable mental representation of semantic relatedness between medical terms that is distinct from similarity and independent of the context in which the terms occur. This study produced a validated publicly available dataset for developing automated approaches to measuring semantic relatedness and similarity.

Towards a framework for developing semantic relatedness reference standards

Our objective is to develop a framework for creating reference standards for functional testing of computerized measures of semantic relatedness. Currently, research on computerized approaches to semantic relatedness between biomedical concepts relies on reference standards created for specific purposes using a variety of methods for their analysis. In most cases, these reference standards are not publicly available and the published information provided in manuscripts that evaluate computerized semantic relatedness measurement approaches is not sufficient to reproduce the results. Our proposed framework is based on the experiences of medical informatics and computational linguistics communities and addresses practical and theoretical issues with creating reference standards for semantic relatedness. We demonstrate the use of the framework on a pilot set of 101 medical term pairs rated for semantic relatedness by 13 medical coding experts. While the reliability of this particular reference standard is in the "moderate" range; we show that using clustering and factor analyses offers a data-driven approach to finding systematic differences among raters and identifying groups of potential outliers. We test two ontology-based measures of relatedness and provide both the reference standard containing individual ratings and the R program used to analyze the ratings as open-source. Currently, these resources are intended to be used to reproduce and compare results of studies involving computerized measures of semantic relatedness. Our framework may be extended to the development of reference standards in other research areas in medical informatics including automatic classification, information retrieval from medical records and vocabulary/ontology development.

Effectiveness of a bed positioning program for treating older adults with knee contractures who are institutionalized

BACKGROUND AND PURPOSE

Although contractures in patients in long-term care institutions are an important issue, there have been only a few studies that have evaluated interventions for contractures. The purpose of this study was to determine the effectiveness of a bed positioning program (BPP) for the treatment of patients with knee flexion contractures.

SUBJECTS

Sixteen patients with a high level of cognitive and functional impairment (mean age=82 years, SD=6.48, range=71-93) in a chronic care hospital participated in the study.

METHODS

The BPP consisted of stretching a patient's knee into extension and then securing and maintaining the position for a period of 40 minutes, 4 times per week. Participants were randomly assigned to 2 groups (n=8 in each group). One group received a BPP for 8 weeks, followed by 8 weeks of no intervention. The other group received the intervention in the reverse order. Once a week, participants were assessed for range of knee extension, knee pain, and skin integrity.

RESULTS

Twelve participants completed the study. There was no improvement in participants' range of knee extension during the intervention period. Overall, there was no difference in mean range of knee extension between the intervention period and the no-intervention period.

CONCLUSION AND DISCUSSION

The results of this study do not support the use of a BPP for treating patients with knee flexion contractures.

Analysis of aneuploidy for chromosomes 13, 21, X and Y by multicolour fluorescence in situ hybridisation (FISH) in a 47,XYY male

The frequency of aneuploid sperm was assessed by fluorescence in situ hybridisation (FISH) in a 47,XYY male previously studied by sperm karyotyping. A total of 20,021 sperm were studied: 10,017 by two-colour FISH for chromosomes 13 and 21 and 10,002 by three-colour FISH for the sex chromosomes using chromosome 1 as an autosomal control for diploidy and lack of hybridisation. Results were compared with more than 500,000 sperm from 18 normal men. The frequencies of X-bearing (49.4%) and Y-bearing sperm (49.8%) were not significantly different from 50% as shown in our sperm karyotyping study. There was no significant increase in the frequency of diploid sperm compared with control donors. There was a significant increase in the frequency of disomy for chromosome 13 (p < 0.0001) and XY disomy (p = 0.0008) compared with control donors. However, since the frequency of disomy was 0.40% for chromosome 13 and 0.55% for XY disomy, it is not surprising that these increases were not discovered previously in our analysis of 75 sperm karyotypes. Our results suggest that the extra Y chromosome is eliminated during spermatogenesis in the majority of cells but that there may be a small but significant increase in the frequency of aneuploid sperm in these men.

Abnormalities for chromosomes 13 and 21 detected in spermatozoa from infertile men

Sperm samples from infertile men with oligozoospermia or teratozoospermia were studied by multicolour fluorescence in-situ hybridization (FISH) using DNA probes for chromosomes 13 and 21. A total of 90 809 sperm nuclei from nine infertile men and 182 799 sperm nuclei from 18 control donors were analysed. There was a highly significant increase in the frequency of spermatozoa disomic for chromosome 13 in infertile patients (0.28%) compared to control donors (0.13%) (two-tailed Z statistic P < 0.0001) and for chromosome 21 (0.48% in infertile men versus 0.37% in controls, P < 0.0001). Also there was a significantly increased frequency of diploid spermatozoa in infertile men (0.85%) compared to control donors (0.66%) (P < 0.0001). Our previous studies on these same infertile patients demonstrated increased frequencies of sperm disomy for chromosomes 1 and XY. This suggests that infertile men, who are prime candidates for intracytoplasmic sperm injection, may be at a very small increased risk of aneuploid offspring.

Donor age and the frequency of disomy for chromosomes 1, 13, 21 and structural abnormalities in human spermatozoa using multicolour fluorescence in-situ hybridization

The purpose of this study was to determine if a donor age effect exists for the frequency of aneuploidy and other chromosome abnormalities in human spermatozoa. Sperm samples were collected from 18 healthy men from the general population. Each individual belonged to one of six age groups (20-24, 25-29, 30-34, 35-39, 40-44, > or = 45 years) with three men in each group. Two multicolour fluorescence in-situ hybridizations were performed on spermatozoa from each donor using probes for chromosomes 13 and 21, and two chromosome 1-specific probes allowed for detection of duplications and deletions as well as disomy of chromosome 1. The abnormality frequencies and the Pearson correlation coefficients were calculated to determine if a relationship existed between donor age and the frequency of chromosome abnormalities in spermatozoa. A statistically significant association with donor age was detected for the frequency of acentric fragments of chromosome 1 (P < 0.05).

A Pro504 --> Ser substitution in the beta-subunit of beta-hexosaminidase A inhibits alpha-subunit hydrolysis of GM2 ganglioside, resulting in chronic Sandhoff disease

The GM2 gangliosidoses are caused by mutations in the genes encoding the alpha- (Tay-Sachs) or beta- (Sandhoff) subunits of heterodimeric beta-hexosaminidase A (Hex A), or the GM2 activator protein (AB variant), a substrate-specific co-factor for Hex A. Although the active site associated with the hydrolysis of GM2 ganglioside, as well as part of the binding site for the ganglioside-activator complex, is associated with the alpha-subunit, elements of the beta-subunit are also involved. Missense mutations in these genes normally result in the mutant protein being retained in the endoplasmic reticulum and degraded. The mutations associated with the B1-variant of Tay-Sachs are rare exceptions that directly affect residues in the alpha-active site. We have previously reported two sisters with chronic Sandhoff disease who were heterozygous for the common HEXB deletion allele. Cells from these patients had higher than expected levels of mature beta-protein and residual Hex A activity, approximately 20%. We now identify these patients' second mutant allele as a C1510T transition encoding a beta-Pro504 --> Ser substitution. Biochemical characterization of Hex A from both patient cells and cotransfected CHO cells demonstrated that this substitution (a) decreases the level of heterodimer transport out of the endoplasmic reticulum by approximately 45%, (b) lowers its heat stability, (c) does not affect its Km for neutral or charged artificial substrates, and (d) lowers the ratio of units of ganglioside/units of artificial substrate hydrolyzed by a factor of 3. We concluded that the beta-Pro504 --> Ser mutation directly affects the ability of Hex A to hydrolyze its natural substrate but not its artificial substrates. The effect of the mutation on ganglioside hydrolysis, combined with its effect on intracellular transport, produces chronic Sandhoff disease.

Transgenic expression of a CD46 (membrane cofactor protein) minigene: studies of xenotransplantation and measles virus infection

CD46 (membrane cofactor protein) is a human cell-surface regulator of activated complement and a receptor for the measles virus. A CD46 transgenic mouse line with an expression pattern similar to that of human tissues has been produced, to develop an animal model of (i) the control of complement activation by complement regulators in hyperacute rejection of xenografts, and (ii) measles virus infection. The mouse line was made using a CD46 minigene that includes promoter sequence and the first two introns of genomic CD46, which was coinjected into mouse ova with chicken lysozyme matrix attachment region DNA. A high level of CD46 expression in homozygotic transgenic mice was obtained with spleen cells having approximately 75% of the level found on human peripheral blood mononuclear cells. CD46 was detected in all tissues examined by immunohistochemistry, radioimmunoassay and Western blotting, showing that these mice were suitable for transplantation and measles virus infection studies. It also indicated that the transgene included the important regulatory elements of the CD46 promoter. Transgenic spleen cells were significantly protected in vitro from human complement activated by either the classical or alternative pathways and from alternative pathway rat complement. Furthermore, transgenic mouse hearts transplanted to rats regulated complement deposition in an in vivo model of antibody-dependent hyperacute xenograft rejection. Similar to human lymphocytes, transgenic lymphoblasts could be infected in vitro with measles virus; infected cells expressed viral proteins and produced infectious viral particles. The data demonstrate the suitability of this minigene for obtaining high-level CD46 expression sufficient for enhanced resistance of transgenic cells to complement attack and for obtaining wide tissue distribution of CD46, analogous to human tissues and, therefore, useful for comparative studies.

Genotype-phenotype pitfalls in Gaucher disease

Gaucher disease (GD), caused by inherited deficiency of beta-glucocerebrosidase (beta-Glc, EC 3.1.2.45), is classified type I if the CNS is not involved (non-neuronopathic), type II if CNS involvement is early and rapidly progressive (acute neuronopathic), and type III if CNS involvement occurs later and is slowly progressive (subacute neuronopathic). The clinical course is not predictable by measurement of residual beta-Glc activity. Patient classification by identification of specific mutations is more promising: homozygosity for the common A5841->G (N370S) mutation invariably predicts type I; homozygosity for the T6433->C (L444P) mutation usually indicates type III (Norbottnian). Type II disease patients often carry the T6433->C allele together with a complex allele derived in part from the downstream pseudogene by crossover or gene conversion, producing a T6433->C substitution, plus 2 or 3 additional single base substitutions (fusion gene). Employing selective PCR amplification of the structural gene, we detected homozygous T6433C (L444P) point mutations in a Caucasian boy, initially classified as having GD type I, who succumbed to severe visceral GD before age 3 years. A second novel PCR procedure for discriminating between the normal gene and the fusion gene confirmed the homozygous point mutation results. Post mortem neuropathological findings showed neuronal complex lipid accumulation consistent with late-onset type III disease. Although in Norbottnian patients it is generally accepted that onset of neurological findings is delayed, patients with the L444P/L444P genotype can only be initially classified as type III with this ancestry. Other patients described sporadically elsewhere are invariably considered type I until neurological findings arise.(ABSTRACT TRUNCATED AT 250 WORDS)

Identification of a processed pseudogene related to the functional gene encoding the GM2 activator protein: localization of the pseudogene to human chromosome 3 and the functional gene to human chromosome 5

The GM2 activator protein is an essential substrate cofactor for the hydrolysis of GM2 ganglioside by lysosomal beta-hexosaminidase A (EC 3.2.1.52). There have been conflicting reports as to the chromosomal localization of the gene encoding the activator. We demonstrate here that these conflicts were caused by the presence of a previously unidentified processed activator-pseudogene on chromosome 3, and we confirm a previous ELISA-based localization of the functional activator gene to chromosome 5. Our data indicate that the functional activator locus can still be considered a candidate site for defects causing some forms of spinal muscular atrophy.

Characterization of two HEXB gene mutations in Argentinean patients with Sandhoff disease

Beta-hexosaminidase A (beta-N-acetyl-D-hexosaminidase, EC 3.2.1.5.2) is a lysosomal hydrolase composed of an alpha- and a beta-subunit. It is responsible for the degradation of GM2 ganglioside. Mutations in the HEXB gene encoded beta-subunit cause a form of GM2 gangliosidosis known as Sandhoff disease. Although this is a rare disease in the general population, several geographically isolated groups have a high carrier frequency. Most notably, a 1 in 16-29 carrier frequency has been reported for an Argentinean population living in an area contained within a 375-km radius from Córdoba. Analysis of the genomic DNA of two patients from this region revealed that one was homozygous for a G to A substitution at the 5' donor splice site of intron 2. This mutation completely abolishes normal mRNA splicing. The other patient was a compared of the intron 2 G-->A substitution and a second allele due to a 4-bp deletion in exon 7. The beta-subunit mRNA of this allele is unstable, presumably as a result of an early stop codon introduced by the deletion. Two novel PCR-based assays were developed to detect these mutations. We suggest that one of these assays could be modified and used as a rapid screening procedure for 5' donor splice site defects in other genes. These results provide a further example of the genetic heterogeneity that can exist even in a small geographically isolated population.

An unusual splicing mutation in the HEXB gene is associated with dramatically different phenotypes in patients from different racial backgrounds

Sandhoff disease is caused by mutations affecting the beta subunit of lysosomal beta-hexosaminidase (EC 3.2.1.52) and displays a wide spectrum of clinical phenotypes. We report a 57-year-old patient with a very mild phenotype, although residual hexosaminidase A activity in his cultured fibroblasts was less than 3% of normal activity, a level observed in juvenile onset patients. Northern and Western blot analyses confirmed a similar low level of beta subunit-mRNA and mature beta-protein, respectively. Two mutations of the HEXB gene were identified in this patient, a partial 5' gene deletion (a null allele), and a C----T transition 8 nucleotides downstream from the intron 10/exon 11 junction affecting the splicing of the beta subunit-mRNA. In their homozygous forms, the 5' deletion has been previously shown to result in a severe infantile phenotype, and the C----T transition in a juvenile phenotype. The genotype and the low level of residual hexosaminidase A activity would be expected to produce a juvenile Sandhoff phenotype in this patient, as well as in four of his six clinically normal siblings. The biochemical basis of his mild phenotype is uncertain, but may result from genetic variations in the RNA splicing machinery.

Two small deletion mutations of the HEXB gene are present in DNA from a patient with infantile Sandhoff disease

Lysosomal beta-hexosaminidase (EC 3.2.1.52) occurs as two major isozymes hexosaminidase A (alpha beta) and B (beta beta). The alpha subunit is encoded by the HEXA gene and the beta subunit by HEXB gene. Defects in the alpha or beta subunits lead to Tay-Sachs or Sandhoff disease, respectively. While many HEXA gene mutations have been reported only three HEXB gene mutations are known. We report the characterization of two rare HEXB mutations present in genomic DNA from a single fibroblast cell line, GM203, taken from a patient with the infantile form of Sandhoff disease. The first is a single base pair deletion in exon 7 changing the codon for Gly-258, GGA, to GA and the second, a two base pair deletion in exon 11 changes the codons for Arg-435/Val-436, AGA/GTC, to AGTC. Each mutation produces a frame shift in the affected allele that results in a premature stop codon 17 or 20 codons downstream, respectively. These mutations also result in the inability to detect beta-mRNA by Northern blot analysis of total mRNA. These data are consistent with the idea that the severe infantile form of Tay-Sachs or Sandhoff disease is associated with a total lack of residual hexosaminidase A activity.

Isolation and expression of a full-length cDNA encoding the human GM2 activator protein

We report the construction of a cDNA clone encoding a functional GM2-activator protein. The sequence of the complete 5' end of the coding region was determined by direct nucleotide sequencing of a fragment generated by multiple RACE PCR procedures from Hela cell cDNA. Specific oligonucleotides were synthesized from these data which allowed us to produce a PCR fragment that contained the complete coding sequence of the protein. This was then cloned into a mammalian expression vector. The ability of purified hexosaminidase A (beta-N-acetylhexosaminidase, EC 3.2.1.52) to hydrolyse labeled GM2 ganglioside was enhanced 10-fold more by the addition in the assay mix of lysate from transfected COS-1 cells than by the addition of identical amounts of lysate from mock transfected cells. Direct sequencing of PCR fragments from two sources also identified three polymorphisms.

Structure-function studies of human interferons-alpha: enhanced activity on human and murine cells

To identify functionally important regions of the human interferon (IFN)-alpha molecule, mutagenesis in vitro of human IFN-a genes was used to create analogs with deletions or specific amino acid replacements. These analogs were expressed in vitro using SP6 RNA polymerase and a rabbit reticulocyte lysate protein synthesis system. Deletion of 7 highly conserved hydrophilic amino acids from the C-terminus of human IFN-alpha 4 reduced, but did not abolish, antiviral activity on human cells. However, analogs with deletions of 15 or 25 amino acids from the C-terminus, or 28 amino acids from the N-terminus, had no measurable antiviral activity. The antiviral activity of human IFN-alpha 4 was increased by substitution of cysteine for serine at position 86, and lysine for arginine at position 121. However, other amino acid substitutions at positions 121, 122 or 123 reduced antiviral activity. The size of the side chain of the amino acid residue at position 130 was shown to be important. Replacement of the absolutely conserved leucine residue at position 131 with glutamine had little effect on antiviral activity. However, the introduction of a proline residue at this position abolished antiviral activity, probably due to the formation of a beta turn in the polypeptide chain. The antiviral activity of human IFN-alpha 4 on murine cells was increased by substitutions at positions 86, 121 and 133. This study illustrates the utility of the in vitro mutagenesis and rabbit reticulocyte lysate systems for the investigation of structure-function relationships, and extends our knowledge of the biologically active regions and species specificity of the human IFN-alpha molecule.

Structure and distribution of an Alu-type deletion mutation in Sandhoff disease

Sandhoff disease is a recessively inherited lysosomal storage disease resulting from a deficiency of beta-hexosaminidase activity. The enzyme occurs in two major forms, beta-hexosaminidase A, composed of an alpha- and beta-subunit and beta-hexosaminidase B, composed of two beta-subunits. Both isozyme activities are deficient in Sandhoff disease, owing to mutations of the HEXB gene encoding the common beta-subunit. We have cloned and fully characterized a deletion at the HEXB gene from fibroblasts of a patient with the infantile form of Sandhoff disease. The deletion removes approximately 16 kb of DNA including the HEXB promoter, exons 1-5 and part of intron 5. It most likely arose from recombination between two Alu sequences, with the breakpoints occurring at the midpoint between the left and right arms in each case and regenerating an intact Alu element in the deletion sequence. The deletion allele accounts for 27% of the Sandhoff mutant alleles we analyzed. Two cell lines were shown to be homozygous for the deletion and both had the infantile form of the disease. Four additional patients were compound heterozygotes with other mutations, all of whom displayed a different clinical phenotype. Finally, the mutant allele was present in different ethnic backgrounds, suggesting that it may have been subject to genetic drift.

Structure-function studies of interferon-alpha based on random mutagenesis and expression in vitro

An efficient procedure for random chemical mutagenesis was used to create analogs of human interferon (IFN)-alpha 4. Unique restriction enzyme sites were introduced into the human IFN-alpha 4 gene to enable cassetting of the gene for localized random mutagenesis. Single-stranded IFN-alpha 4 DNA was treated with nitrous acid, followed by second-strand synthesis using reverse transcriptase. A 72 base pair cassette spanning the coding region for amino acid residues 120 to 136 (120-136 region) was isolated and cloned into a phagemid vector adjacent to a GC-rich sequence. A DNA segment comprising the IFN-alpha 4 cassette sequence and the GC clamp was excised and electrophoresed on a denaturing gradient gel, which allowed the separation from unmutated DNA of DNA fragments with single base pair changes. DNA fragments with mobility different from that of the unmutated fragment were pooled and cloned into an expression vector. Using this procedure, mutations were found in the DNA of 48% of the clones analyzed. However, mutations at two "hot spots" accounted for 89% of these clones. Four of the IFN-alpha 4 analogs with mutations in the 120-136 region were expressed in vitro. The antiproliferative activities on human Daudi cells of most of the analogs were less than 0.2% of the activity of unmodified IFN-alpha 4, suggesting that the integrity of the carboxy terminus is important for the antiproliferative activity of human IFN-alpha 4.

Functional significance of amino acid residues within conserved hydrophilic regions in human interferons-alpha

Site-directed in vitro mutagenesis was used to create analogs of human interferons (IFNs)-alpha 1 and -alpha 4. Analogs were expressed in vitro using SP6 RNA polymerase and a rabbit reticulocyte lysate cell-free protein synthesis system. Amino acid substitutions for the highly conserved residues at positions 33, 121, 122 and 123 greatly reduced the antiviral and antiproliferative activities on human cells of IFNs-alpha 1 and -alpha 4. In general, the amino acid substitutions had much less effect on the antiviral activities on bovine, compared with human, cells. Substitutions at positions 31, 41, 42, 124, 134, 135 and 136 had little or no effect on the biological activities of the IFN analogs. The abrogation of antiviral activity resulting from amino acid substitutions for the arginine residue at position 33 suggests that this arginine residue is required for binding to the IFN-alpha receptor on the cell surface.

Structure-function studies of interferons-alpha: amino acid substitutions at the conserved residue tyrosine 123 in human interferon-alpha 1

Analogs of human interferon-alpha 1 (IFN-alpha 1) were created in vitro by site-directed mutagenesis to investigate the structural requirements at amino acid position 123 for binding to the IFN receptor, antiviral activity, and antiproliferative activity. The tyrosine residue 123, which is conserved in all known mammalian IFNs-alpha and -beta, was replaced by each of 6 amino acids or was deleted from the protein. Several of the substitutions at position 123 partly or completely abrogated antiviral and antiproliferative activities of human IFN-alpha 1 when human or murine cells were used but not when bovine cells were used. However, with analogs in which amino acids structurally related to tyrosine, phenylalanine, or tryptophan were substituted at position 123, there was retention of antiviral and antiproliferative activities using homologous cells. Thus, although there is not an absolute requirement for tyrosine at position 123, conformational changes associated with alterations of this residue are prejudicial to the biological functions of the IFN-alpha molecule.

Amino acid substitutions which alter the antiviral activity of human interferon-alpha 1 on mouse cells

Human interferon-alpha 1 (IFN-alpha 1) is one of only three human (Hu) IFN-alphas having significant antiviral activity on mouse cells. Specific amino acid substitutions in HuIFN-alpha 1 in the region from amino acids 121 to 136 indicate that this region is critical to the determination of mouse and human cell antiviral activities. Bovine cell activities are relatively unaffected by changes in this region. In particular we have identified the arginine residue at position 125 of human IFN-alpha 1 as a major mediator of the molecules antiviral activity on mouse cells. Various substitutions in the carboxy-terminal region of human IFN-alpha 1 are also evaluated and discussed in the context of recently published data.

Efficient in vitro expression of interferon alpha analogs using SP6 polymerase and rabbit reticulocyte lysate

We have investigated the use of in vitro expression as a quick and convenient means of screening large numbers of interferon (IFN) analogs generated using in vitro mutagenesis. The IFN-alpha 1 mRNA generated from DNA template using SP6 RNA polymerase is efficiently translated in rabbit reticulocyte lysate (RRL). The antiviral specific activity of this RRL-synthesized IFN-alpha l is equivalent to the yeast-synthesized protein. In contrast with the yeast-expression system, where some IFN-alpha analogs are poorly expressed, all analogs tested were well expressed in RRL.