Flanking Genomic Sequences Research Articles

An insertional mutagenesis approach to Dictyostelium cell death.

Programmed cell death (PCD) in Dictyostelium shows a pattern of ordered degeneration similar to that observed in higher eukaryotes but somewhat different from the most studied form of PCD, i.e. apoptosis. To contribute to a genetic definition of this process, Dictyostelium HMX44A cells have been subjected to insertional mutagenesis, followed by selection based on several rounds of differentiation/regrowth to recover only cells resistant to death. We describe here the approach used, a partial characterization of the first mutant thus obtained called C5 showing some dissociation of cell death signs, and, in this case where plasmid rescue was not possible, as a first step towards identification of the gene at play recovery of genomic flanking sequences via genomic recircularization and PCR. This work demonstrates the feasibility of an insertional mutagenesis approach to obtain death-resistant mutants in Dictyostelium.

Genomic Organization of the MouseReelinGene

Reelin is the protein defective inreelermice, an extensively studied model of brain development. Thereelingene (symbolReln) codes for a protein of the extracellular matrix that contains eight successive repeats of 350 to 390 amino acids. In this work, we describe the genomic structure of the mousereelingene and the 5′-flanking genomic DNA sequences. Thereelingene is composed of 65 exons spread over approximately 450 kb of genomic DNA. We identified differentreelintranscripts, formed by alternative splicing of a microexon as well as by use of two different polyadenylation sites. All splice sites conform to the GT-AG rule, except for the splice donor site of intron 30, which is GC instead of GT. A processed pseudogene is present in intron 42. Its nucleotide sequence is 86% identical to the sequence of the rat RDJ1 cDNA, which codes for a DnaJ-like protein of the Hsp40 family. Comparison of 8 intron positions in mouse and humanreelingenes reveals a highly conserved genomic structure, suggesting a similar structure of the whole gene in both species. We identified two transcription start sites embedded within a CpG. The promoter region contains putative recognition sites for the transcription factors Sp1 and AP2 but lacks TATA and CAAT boxes. The presence of tandemly repeated regions in the Reelin protein suggests that gene duplication events occurred during evolution. By comparison of the amino acid sequences of the eight repeats and the positions of introns, we suggest a model for the evolution of the repeat coding portion of thereelingene from a putative ancestral minigene.

Nonrandom location of IS1 elements in the genomes of natural isolates of Escherichia coli.

We have studied the spatial distribution of IS1 elements in the genomes of natural isolates comprising the ECOR reference collection of Escherichia coli. We find evidence for nonrandomness at three levels. Many pairs of IS1 elements are in much closer proximity (< 10 kb) than can be accounted for by chance. IS1 elements in close proximity were identified by long-range PCR amplification of the genomic sequence between them. Each amplified region was sequenced and its map location determined by database screening of DNA hybridization. Among the ECOR strains with at least two IS1 elements, 54% had one or more pairs of elements separated by < 10 kb. We propose that this type of clustering is a result of "local hopping," in which we assume that a significant proportion of tranposition events leads to the insertion of a daughter IS element in the vicinity of the parental element. A second level of nonrandomness is found in strains with a modest number of IS1 elements that are mapped through the use of inverse PCR to amplify flanking genomic sequences: in these strains, the insertion sites tend to be clustered over a smaller region of chromosome than would be expected by chance. A third level of nonrandomness is observed in the composite distribution of IS elements across strains: among 20 mapped IS1 elements, none were found in the region of 48-77 minutes, a significant gap. One region of the E. coli chromosome, at 98 min, had a cluster of IS1 elements in seven ECOR strains of diverse phylogenetic origin. We deduce from sequence analysis that this pattern of distribution is a result of initial insertion in the most recent common ancestor of these strains and therefore not a hot spot of insertion. Analysis using long-range PCR with primers for IS2 and IS3 also yielded pairs of elements in close proximity, suggesting that these elements may also occasionally transpose by local hopping.

Promoter analysis in living zebrafish embryos identifies a cis-acting motif required for neuronal expression of GATA-2.

We have used zebrafish embryos to dissect the promoter activity of a gene with a complex expression pattern during embryogenesis. GATA-2 is a transcription factor required for hematopoiesis and is dynamically expressed in hematopoietic tissues and in the central nervous system. Using constructs containing zebrafish GATA-2 genomic flanking sequences and the green fluorescent protein (GFP) reporter gene, we demonstrate that distinct regulatory domains are required for hematopoietic, enveloping layer (EVL), and neuronal expression of GATA-2. During gastrulation, GFP expression is confined to the ventral ectoderm and lateral mesoderm and is lacking in the dorsal shield. Cells derived from the regions expressing GFP give rise to hematopoietic progenitors, EVL cells, and neurons. Deletion analysis of the 7.3-kb GATA-2 promoter region revealed that a 1.1-kb DNA sequence is critical for expression of GATA-2 in neurons. Fine mapping revealed that a 31-bp region is required for neuron enhancer activity, and mutagenesis showed that the DNA motif CCCTCCT is essential for GATA-2 promoter activity in the central nervous system of zebrafish. Our use of zebrafish embryos can be exploited as a whole animal system for the dissection of any developmentally regulated vertebrate promoter.

Cytokines are increased in the rat hippocampus after serotonergic neuron degeneration and upregulate the expression of GDH, an enzyme involved in glutamate detoxification

The degeneration of serotonergic neurons increases the expression of glutamate dehydrogenase (GDH) in hippocampal astrocytes. This process was demonstrated to be independent of the serotonin level. At the same time, upregulation of tumor necrosis factor (TNF) a and interleukin (IL)-1α mRNA were observed, whereas levels of transforming growth factor (TGF) β1 mRNA remained unchanged. The level of GDH mRNA was increased in primary cultures of hippocampal astrocytes treated with TNFα and IL-1α suggesting that these cytokines act on the GDH metabolism. TNFa and IL-1α induced an increase in GDH promoter activity in C8S (an astrocytic cell line) transfected with constructs containing 5′ flanking genomic sequences of GDH driving the expression of a reporter gene. These observations suggest that cytokines may be signals that upregulate the astrocytic GDH expression in response to the degeneration of serotonergic terminals in the hippocampus.

A new method for the identification and isolation of genes essential for Arabidopsis thaliana seed germination.

Seed viability, dormancy and germination efficiency are very important aspects of the life cycle of plants and their potential to survive and spread in the environment. To characterize the genes controlling these processes, we have devised a technique for the selection of mutants impaired in seed germination. Selection for such a trait is complicated by physiological factors that interact with these processes and affect seed germination efficiency. The distinction between low seed germination potential due to physiological factors that interfere with seed maturation or germination and germination deficiency due to genetic factors was based on screening for tagged mutations. Arabidopsis thaliana T-DNA primary transformants obtained by an in planta transformation technique are all heterozygotes. We screened for lack of germination of 1/4 of the seeds in the progeny of independent transformants, and simultaneously for the abnormal segregation (2:1 instead of 3:1) of a kanamycin resistance marker carried by the T-DNA inserted into the genome of these primary transformants in the plants that germinate. This yielded several mutants affected in the germination processes. One of the mutants, designated ABC33, was further characterized. Once the viable embryos from non-germinating seeds were removed from their testa, they grew and displayed a dwarf phenotype which could be complemented by providing gibberellic acid. A genetic and molecular analysis, based on the characterization of the flanking genomic sequences of the T-DNA insert, showed that ABC33 is a new loss-of-function allele at the GA1 locus.

Genomic Organization and 5′-Flanking DNA Sequence of the Murine Stomatin Gene (Epb72)

Stomatin is a poorly understood integral membrane protein that is absent from the erythrocyte membranes of many patients with hereditary stomatocytosis. This report describes the cloning of the murine stomatin chromosomal gene, determination of its genomic structure, and characterization of the 5′-flanking genomic DNA sequences. The stomatin gene is encoded by seven exons spread over ∼25 kb of genomic DNA. There is no concordance between the exon structure of the stomatin gene and the locations of three domains predicted on the basis of protein structure. Inspection of the 5′-flanking DNA sequences reveals features of a TATA-less housekeeping gene promoter and consensus sequences for a number of potential DNA-binding proteins.

Genomic structure of a Y-specific ribonucleic acid binding motif-containing gene: a putative candidate for a subset of male infertility.

The genetic basis of infertility remains unclear in a majority of infertile men. Deletion mapping studies suggest that genes on the long arm of the Y-chromosome (Yq) may be important in the spermatogenic process and may play a pathogenetic role in a subset of infertile men. Complementary DNA sequences of two Y-specific genes that contain ribonucleic acid binding motifs and, therefore, referred to as RBM genes (previously named YRRM) were published recently. To develop a PCR-single strand conformation polymorphism strategy for detection of point mutations in the RBM gene(s) in infertile men, we determined the genomic structure and flanking sequences at the intron-exon junctions. Two separate strategies were used in parallel to isolate the genomic fragment bearing the RBM gene. The first strategy employed screening of a P1 genomic library using PCR primers corresponding to the sequences in the 5'- and 3'-ends of the published RBM-1 complementary DNA sequence. The second strategy used subcloning of the YAC clone 925D10 (that contained the RBM gene described here) into cosmids. The P1 and cosmid clones were further restriction mapped and subcloned for DNA sequencing. Because the sequences contained in the P1 and cosmid clones were identical, the sequence information was pooled. A 15-kilobase genomic segment includes the entire RBM gene. The genomic structure of this RBM gene is characterized by 12 exons and 11 introns. There is considerable homology among exons VII, VIII, IX, and X; each encodes one of the SRGY boxes. Several introns also have a high degree of homology among them (introns VI, VII, VIII, and IX). Eleven of the 12 exons have complete sequence homology with the RBM-1 sequence. There is 1 base difference in exon IV at position 495 (a T in the previously published DNA sequence vs. an A in the sequence reported here). The exonic sequences of this gene are distinct from that of the RBM-2 gene. The flanking sequences at the exon-intron junctions were also determined and are reported. Reverse transcription-PCR analysis, using human testis ribonucleic acid suggests that this gene is either not expressed in the testis or, more likely, the single base difference from RBM1 represents a polymorphism in the YAC clone. A high degree of homology between intronic and exonic sequences within the same gene and between different members of the RBM gene family (data not reported in this paper) suggests origin from common ancestral sequences; this also indicates that development of a single strand conformation polymorphism approach for detection of point mutations is likely to prove difficult for some of the exons of this gene.

Purification of Nuclear Proteins from Human HeLa Cells That Bind Specifically to the Unstable Tandem Repeat (CGG)n in the Human FMR1 Gene

Autonomous expansions of trinucleotide repeats with the general structure 5'-d(CNG)n-3' are associated with several human genetic diseases. We have characterized nuclear proteins binding to the unstable 5'-d(CGG)n-3' repeat. Its expansion in the human FMR1 gene leads to the fragile X syndrome, one of the most frequent causes of mental retardation in human males. Electrophoretic mobility shift assays using nuclear extracts from several human and other mammalian cell lines and from primary human cells demonstrated specific binding to double-stranded DNA fragments containing only a 5'-d(CGG)17-3' repeat or the repeat and flanking genomic sequences of the human FMR1 gene. Protein binding was inhibited by complete methylation of the trinucleotide repeat. The complex formed with crude nuclear extract apparently did not contain the human transcription factor Sp1 that binds to a characteristic GC-rich sequence. A 20-kDa protein involved in specific binding to the double-stranded 5'-d(CGG)17-3' repeat was purified from HeLa nuclear extracts by DNA affinity chromatography.

Human Protein Tyrosine Phosphatase 1C (PTPN6) Gene Structure: Alternate Promoter Usage and Exon Skipping Generate Multiple Transcripts

PTPN6, a protein tyrosine phosphatase with two Src homology-2 domains, is expressed predominantly in cells of the hematopoietic lineage. In addition, cDNAs encoding isoforms of the enzyme differing from the latter at their N-terminal have been detected in normal and transformed epithelial cells. To understand further the regulation and expression of the PTPN6 gene, we have isolated and characterized overlapping human genomic lambda clones containing the entire locus. The gene, located on chromosome 12, consists of 17 exons spanning 17 kb of DNA. Three nonhematopoietic PTPN6 transcripts were identified in a variety of cell lines and were shown to be transcribed from a common promoter. Alternate splicing and exon skipping result in transcripts encoding the two SH2 domains and minor transcripts from which part of or the entire N-terminal SH2 domain coding sequences are missing. The hematopoietic form of PTPN6 transcript is initiated at a downstream promoter separated by 7 kb from the first. This promoter is active exclusively in cells of the hematopoietic lineage. Characterization of the 5' ends of the PTPN6 mRNAs by RT-PCR and analysis of the flanking genomic sequences identified putative initiation sites within the two promoters.

Functional identification of the promoter for the gene encoding the alpha subunit of calcium/calmodulin-dependent protein kinase II.

To examine the expression of the alpha subunit of calcium/calmodulin-dependent protein kinase II, various 5' flanking genomic sequences were inserted into a chloramphenicol acetyltransferase (CAT) reporter plasmid and CAT enzyme activities were analyzed in transfected NB2a neuroblastoma cells and mRNA transcription was analyzed by nuclease protection assays. A core promoter was identified which contained an essential TATA element located 162 nt 5' to the transcription start site. Sequences 3' to the transcription start site, as well as 5' to the TATA element, increased levels of CAT activity in transfected cells. The alpha-subunit gene promoter displayed higher CAT activities, relative to a simian virus 40 promoter, in transfected neuronal cell lines than in nonneuronal cell lines. Results also suggested that sequence surrounding the natural alpha-gene transcription initiation site may be important for targeting transcription initiation 162 nt downstream of its TATA element.

Trapping of genes induced upon growth arrest after treatment with antiestrogen or retinoids using retroviral promoter trap

Although chemopreventive anti-steroids such as the antiestrogens are thought to act through competitive inhibition of agonist binding to estrogen receptors, it has been postulated that the estrogen receptor changes its conformation when bound to a strong antiestrogen such as ICI-164,384. We hypothesized that such conformationally changed receptors could bind specific recognition sequences in the genome and activate specific genes that might be involved in growth arrest. In order to identify such genes with a functional assay, we used a retroviral gene trap U3lacZ. We have now isolated MCF-7 breast cancer cell line clones in which the lacZ reporter gene is inserted into the genes activated by either ICI-164,384 or retinoic acid. One such clone, B4, was further characterized. In B4, lacZ activity is induced by ICI-164,384 or trans-retinoic acid, and repressed after treatment with estradiol. Cloning of the 5′-flanking genomic sequence in this clone will be possible using polymerase chain reaction.

Nucleotide sequence of bup, an upstream gene in the bmi-1 proviral insertion locus.

The ability of Moloney murine leukemia virus to accelerate lymphomagenesis in E mu-myc transgenic mice is frequently associated with proviral integration within a locus denoted bmi-1. This locus contains not only the bmi-1 gene implicated as a collaborator with myc in lymphomagenesis but also just upstream an unknown gene denoted bup. The nucleotide sequence reported here for bup cDNA and flanking genomic sequences reveals that this widely expressed gene comprises at least 7 exons and potentially encodes a polypeptide of 195 amino acid residues. Computer searches with this polypeptide sequence revealed no close homolog in the databases, nor any conserved motifs, and it is unrelated to the product of the mel-13 gene, which lies just upstream from the bmi-1 homolog mel-18.

Intron requirement for expression of the human purine nucleoside phosphorylase gene.

Abbreviated purine nucleoside phosphorylase (PNP) genes were engineered to determine the effect of introns on human PNP gene expression. PNP minigenes containing the first intron (complete or shortened from 2.9 kb down to 855 bp), the first two introns or all five PNP introns resulted in substantial human PNP isozyme expression after transient transfection of murine NIH 3T3 cells. Low level human PNP activity was observed after transfection with a PNP minigene containing the last three introns. An intronless PNP minigene construct containing the PNP cDNA fused to genomic flanking sequences resulted in undetectable human PNP activity. Heterogeneous, stable NIH 3T3 transfectants of intron-containing PNP minigenes (verified by Southern analysis), expressed high levels of PNP activity and contained appropriately processed 1.7 kb message visualized by northern analysis. Stable transfectants of the intronless PNP minigene (40-45 copies per haploid genome) contained no detectable human PNP isozyme or mRNA. Insertion of the 855 bp shortened intron 1 sequence in either orientation upstream or downstream of a chimeric PNP promoter-bacterial chloramphenicol acetyltransferase (CAT) gene resulted in a several-fold increase in CAT expression in comparison with the parental PNP-CAT construct. We conclude that human PNP gene expression at the mRNA and protein level is dependent on the presence of intronic sequences and that the level of PNP expression varies directly with the number of introns included. The disproportionately greatest effect of intron 1 can be explained by the presence of an enhancer-like element retained in the shortened 855 bp intron 1 sequence.

Structural aspects of the human C5 gene. Intron/exon organization, 5'-flanking region features, and characterization of two truncated cDNA clones.

Human C5 cDNA fragments were used to identify five overlapping cosmid clones that spanned the entire C5 gene. Partial sequencing and Southern analysis of the clones were performed to identify intron/exon boundaries and to map intron size. The human C5 gene is 79 kilobases in length and is comprised of 41 exons. Comparison of C5 with the homologous family members C3 and C4 revealed striking similarities in exon size and number. Less, although significant similarities were also observed with the family member alpha 2-macroglobulin. The transcriptional start site for the C5 gene was observed as a doublet at positions 29 and 28 nucleotides upstream of the ATG start codon. The 5'-flanking region of the gene contains sequences homologous with several known responsive elements, including interferon, interleukin-6, glucocorticoid, estrogen, NF-kappa B, and HNF-1. Two previously identified truncated cDNAs, pHC5A and pHC5B, contain 21 and 16 exons, respectively. The last exon in pHC5A, designated exon 21a, is a product of alternative splicing and is not present in the major full-length transcript. Truncation of pHC5A is the result of an alternative polyadenylation signal located in exon 21a. In pHC5B, exon 16 is extended on the 3' end by additional flanking genomic sequence that also contains an alternative polyadenylation signal.

Liver-specific and position-effect expression of a retinol-binding protein- lacZ fusion gene (RBP- lacZ) in transgenic mice

Transgenic mice carrying a liver-specific promoter fused to a nuclear-targeted lacZ reporter gene were generated. Three separate lines of mice showed liver expression in the adult and no expression elsewhere in the animal. These results show that a 1.7 kb 5′-flanking region in the retinol-binding protein gene contains necessary and sufficient transcriptional signals for expression in adult livers. A fourth line (R197) did not express the transgene in the liver; instead, constitute lacZ expression was seen during postimplantation stages of development from Day 9.5 onwards and appeared to be associated with segmented structures including somites, branchial arches, and hindbrain rhombomeres until late fetal stages. The β-galactosidase positive cells in R197 were later seen to give rise to facial and flank musculature, and to other region-specific subpopulations of the jaws, neocortex, and brain stem. Northern blot analysis for the host retinol-binding protein RNA transcript did not show overlapping tissue expression with the reporter gene and suggests that transgenic phenotype seen in segmented embryonic structures of R197 and other extra-hepatic sites is from novel cis-acting transcriptional specificity. RNase protection assays of the R197 mRNA indicate that the lacZ sequences are appropriately transcribed downstream of the RBP canonical TATA box, even though the RBP promoter is itself silent. This result would suggest host flanking sequences with enhancer activity may have either activated the lacZ reporter gene or cooperated with the RBP promoter to create novel region-specific transcription. Breeding experiments have so far failed to produce offsprings that are homozygous for the transgene, and mating of transgenic F1 siblings routinely produce reduced litter sizes. Embryos that are homozygous for the transgene appear to be unable to survive beyond the egg cylinder stages of development. Thus, disruption of the host genome by insertional mutation appears to be manifest at an earlier stage than when position-effect expression of the transgene is first apparent. These experiments demonstrate that the component parts of a transgene may be subject to differential activation or suppression by host genomic flanking sequences and that even a strong, tissue-specific promoter may be overridden by host genes.

Molecular analysis of two PR-1 pseudogenes from tobacco

Two independent PR-1 lambda genomic clones (W38/1 and W38/3) were isolated and characterized from a tobacco (Nicotiana tabacum cv. Wisconsin 38) library. Neither clone is identical to the previously described PR-1 cDNA clones, and both clones carry mutations within the highly conserved PR-1 protein coding region. For example, clone W38/1 has a GAA Glu codon instead of the translation stop codon thus harbouring an open reading frame extended by 16 additional amino acids. Furthermore, both clones display considerable variations in the genomic flanking sequences when compared to the PR-1a gene. In order to test whether the encoded genes are active, their upstream sequences were fused to the E. coli beta-glucuronidase (GUS) reporter gene. While significant GUS activities as compared to the 35S RNA promoter from cauliflower mosaic virus (CaMV) were obtained with the W38/1 and W38/3 sequences in transient gene expression assays, no transcriptional activities could be observed upon stable transformation of the same constructs. In addition, the protein coding region of W38/1 was joined to the CaMV 35S RNA promoter and transgenic tobacco plants were generated. However, neither transcripts nor a protein could be detected deriving from the W38/1 structural gene with this chimaeric construct in the transformants. Taken together, these data indicate that the genes contained in lambda clones W38/1 and W38/3 are not active in planta.

Structure of the 5' flanking region of the gene encoding human parathyroid-hormone-related protein (PTHrP)

We have characterized a human genomic clone that contains the 5' coding and 5' flanking sequences of the human parathyroid hormone-related protein gene ( PTHrP). The 5' end of the gene contains three exons separated by two small introns of 60 and 165 bp, respectively. The coding region of the PTHrP gene exhibits significant structural homology to the human parathyroid hormone gene ( PTH), including the position of at least two introns. However, there is no significant nucleotide sequence homology to the PTH gene within the intragenic region nor in the flanking genomic sequences. The PTHrP gene has been localized, by chromosomal in situ hybridization to bands p11 or p 12, on human chromosome 12. Analysis of the 5'-noncoding DNA reveals a complex, putative regulatory region, with multiple potential transcription start points. Nucleotide sequence analysis shows the position of one consensus TATA sequence, at −514 bp, from the start of translation whereas the other regulatory domain is located at least 1 kb further 5' to this consensus TATA sequence. Evidence from the structure of a number of cDNA clones, as well as S1 nuclease and primer extension studies supports the hypothesis that the PTHrP gene contains at least two mRNA transcription start points that define two putative regulatory domains. The result of expression from these different promoters combined with an alternative splicing event would be to produce multiple forms of PTHrP mRNA that differ in the 5'-untranslated region. This analysis of the human PTHrP gene is the first report of a PTHrP gene for any species.

Unique sequence organization and erythroid cell-specific nuclear factor-binding of mammalian Θ I globin promoters

The theta 1 globin gene is an alpha globin-like gene, and started to diverge from the other members of the alpha globin family 260 million years ago. DNA sequencing and transcriptional analysis indicated that it is functional in erythroid cells of the higher primates, but not in prosimians and rabbit. The theta 1 promoter region of higher primates including man consists of GC-rich sequences characteristic of housekeeping gene promoters, and CCAAT and TATA boxes located further upstream. It is shown here that the housekeeping gene promoter-like region of human theta 1 contains two tandemly arranged, GC-rich motifs (GC-I and GC-II). Of these, GC-II interacts with nuclear factor(s) present in the globin-expressing, erythroleukemia cell line K562, before and after hemin induction. GC-I, however, interacts with nuclear factor(s) only present in hemin-induced K562 cells. These factors are different from previously reported erythroid cell-specific factors, and are not detectable in non-erythroid Hela cells. Furthermore, the sequence of the motif GC-I and its location relative to ATG codon have been conserved among all known mammalian theta 1 globin genes. Finally, and most interestingly, the CCAAT box of theta 1 is contained within a 38 bp internal segment of Alu repeat sequence. Immediately upstream from this CCAAT box-containing Alu repeat segment is a 241 bp Alu repeat pointing in the opposite direction. The conservation of this novel arrangement among the higher primates suggests that an inserted Alu family repeat and its flanking genomic sequence have co-evolved, for at least 30 million years, to provide the canonical CCAAT and TATA promoter elements of the theta 1 globin genes in higher primates.

A gene activated in mouse 3T3 cells by serum growth factors encodes a protein with "zinc finger" sequences.

We have recently identified by cDNA cloning a set of genes that are rapidly activated in mouse 3T3 cells by serum or purified growth factors. Here we report that the cDNA (clone 268) derived from one of these immediate early genes (zif/268) encodes a protein with three tandem "zinc finger" sequences typical of a class of eukaryotic transcription factors. The mRNA of zif/268 is present in many organs and tissues of the mouse and is especially abundant in the brain and thymus tissue. The 5' genomic flanking sequence of zif/268 has sequences related to binding sites for known regulatory proteins, including four sequences that resemble the core of the serum response elements (SREs) upstream of the c-fos and actin genes. The SRE-like sequences could be responsible for the coordinate activation of zif/268 and fos after serum stimulation of 3T3 cells.