Early Nodulation Gene Expression example essay topic

Abstract Leguminous plant roots form endosymbioses with both bacteria and fungi forming nitrogen-fixing root nodules and arbuscular mycorrhiza (AM), respectively. The physiological outcomes of both symbioses are quite dissimilar, however, several studies have shown that nodulation defective mutant are often defective in arbuscular mycorrhiza formation, this is indicative of a common genetic overlap in perception of endosymbiosis signals in the host. Analysis has shown several genes to integral to this common system. Study of common phenotypic markers, such as calcium spiking and early nodulation gene expression, has helped to order the action of the common symbiosis genes. The recent cloning of a novel receptor like-kinase has confirmed its role in transduction of both bacterial and fungal symbiotic signals.

The evolutionary history of AM-host interactions dates back ~450 MYA, and it is thought the more recent rhizobia-legume interaction may have evolved by recruiting plant factors originally used for AM perception and formation. Arbuscular mycorrhiza (AM) is an intracellular plant-fungal symbiosis, leading to a greatly improved uptake of phosphate from the soil, formed between most land plants and the zygomycete fungi belonging in the order of Glo males (Brundrett, 2002). Flavonoids exudates from the root hairs of the host plants cause an increase in AM fungal hyphae (Nair et al., 1991). The direct contact of these hyphae with root hair epidermis stimulates the formation of an appressoria, a highly branched network of swollen hype a.

These fungi grow towards the cortex of the roots upon which they differentiate into specialised branched structures known as arbuscular. The dense network of branches provides the large surface area for ion uptake (for review see Smith and Smith, 1997). Fossil records show AM-like interactions in early land plants ~450 MYA (Remy, 1994), however, intercellular bacterial root nodulation is not detected until much later (Herendeen, 1999). Root nodule symbioses are formed between leguminous plants of the Euros id I clade and a unique diverse group of bacteria called rhizobia.

The Rhizobial detection of plant flavonoids induces the release of nodulation factors (NF), symbiotic signalling compounds identified as lipo chitin oligosaccharides. The basic backbone of all NFs is B-1, 4-linked N-acyl-D-glucosamine four or five units in length, which can be modified at the terminal sugar residue or within the acyl chain (Perret, 2000). Species-specific modifications provide rhizobia with a narrow host range e.g. Sino rhizobium Meliloti can nodulate species of Medicago, Melilotus, and Trigon ella; Meso rhizobium loti can nodulate species of lotus and Lupinus; Rhizobium leguminosarum bv. viciac can form nodulate species of pea (Pisum sativum), Vicia, Lens, and Lathy rus; and Rhizobium leguminosarum bv. trifolii can nodulate species of clover (Trifolium), however there are exceptions, Rhizobium strain NGR 234 was shown to nodulate 232 species of legume and can even nodulate the non-legume Parasponia anderson ii (Perret et al., 2000; Pueppke and Broughton, 1999). Mutational studies into NF production have been shown to alter host specificity and are taken as evidence that plants discriminate between Rhizobia by recognition of their NFs (Perret et al., 2000).

NF signalling molecules induce several accommodating changes in the host. NF detection causes root hair deformation (swelling and branching), membrane depolarisation, extracellular alkalinisation, followed by curling of the root tip and an entrapment of the bacteria (Wegel et al, 1998). Entrapment allows the formation of an infection thread, a host controlled inward growing tubular structure, which descends to the root hair cortex (Albrecht et al., 1999). Clusters of cortical cells differentiate to from a primordial nodule, which is subsequently infected by bacteria released from the infection thread. The infected nodule provides the ideal environment to reduce nitrogen into ammonia for uptake by the host.

Whilst NFs have been shown to act as signal molecules between the rhizobia and host the equivalent signal molecule between AM and host has not yet been discovered. At first glance the two intracellular interactions of AM and rhizobia would appear quite different, both inducing different physiological responses in either a promiscuous or species-specific manner. However, studies on model organisms, Lotus japonicus and Medicago truncatula, show that nodulation defective mutants (Nod-) are often coupled with the inability to form mycorrhizal association (Myc-) (Harrison, 1997). In 21 of 45 Nod− mutants of pea and vetch, di allelic crosses have shown that Myc− and Nod− cannot be uncoupled, demonstrating that both phenotypes are derived from mutations in the same gene, and suggest a common pathway for the two symbioses (Harrison, 1997). This convergence of the two associations has been subject to intense study as it suggests that nodule symbiosis may have arisen in part by the hijacking of genes used for the ancient AM symbiosis.

Symbiotic signalling, a convergence system Genetic study of Myc-/Nod- mutants has identified several genes essential to both fungal and bacterial endosymbioses, these are referred to as SYM genes. These genes include MtDMI 1, MtDMI 2, and MtDMI 3 from M. truncatula; PsSym 8, PsSym 9, PsSym 19, and PsSym 30 from pea (P. sativum); LjSYM 2, LjSYM 3, LjSYM 4, LjSYM 15, LjSYM 23, and LjSYM 30 from L. japonicus; and MaSYM 1, MaSYM 3, Pssym 5 from Melilotus alba (Hirsch et al., 2001). Genetic analysis of the pea nodulation mutant Sparkle-R 25, which is mutated in PsSYM 8, show PsSYM 8 to be essential in inducing PsENOD 5 and PsENOD 12 A, both are early nodulin genes required for microbe symbiosis (Albrecht et al., 1998). PsSym 9 and PsSym 30 have also been shown to be vital for root branching deformation (Walker et al., 2000). Study of M. truncatula mutants in three genes MtDMI 1, MtDMI 2, and MtDMI 3 demonstrate their role in inducing early nodulation genes and root hair deformation in response to both AM fungal and NF inoculation (Catoira et al, 2000).

These studies indicate that SYM genes are involved in perception and transduction of AM fungal factor and rhizobia nodulation factor (NF) signals Several common genes induced during both symbioses interactions have been identified, expressed in the epidermis, cortex, and pericycle of the root; ENOD 2, ENOD 5, ENOD 11, ENOD 12, ENOD 40, and rip 1, are all activated upon treatment of NF onto L. japonicus roots and infection by AM fungi, these genes are termed the early nodulin genes (Albrecht et al., 1998; Van Rhijn et al., 1997). Furthermore the leg haemoglobin (LB) gene is activated with 24 to 48 hours of incubation with AM or Nod. LB is thought to be an oxygen buffer in the infection zone, however no oxygen restriction is detected in AM so LB may play an additional role (Strake et a., 2002). One of the earliest known responses to signalling in symbiosis is the establishment of regular periodic calcium spikes.

Addition of NF from R. leguminosarum bv. viciac onto pea root hair induced an increase of calcium within 1-2 minutes, followed 10 minutes later by periodic spikes in calcium of ~200 nM every minute (Ehrhardt et al., 1996). Very little is known about the role of calcium spiking in plants but it's detection early in both AM infection and NF inoculation have placed it early in the common signalling leading to endosymbiotic infection, this combined with expression studies on early nodulin genes has helped to develop a pecking order into the action of the SYM genes. Genetic hierarchy of SYM signalling Mutations in several SYM genes, LjSYM 2, LjSYM 4 from L. japonicus, MtDMI 1, MtDMI 2 from M. truncatula, PsSYM 8, PsSYM 19 from pea, MsN ORK from Medicago sativa, are all Myc-/Nod- and all abolish calcium spiking in presence of NF or AM inoculation, and as a result these genes are placed upstream of this earliest known common response. However, plant lines possessing mutations in these SYM genes still display root swelling in response to NFs (Catoira et al, 2000; Stracke et al., 2002). Taken at face value, this indicates the common pathway to endosymbiotic infection is independent from pathways leading to NF induced root hair swelling (Fig. 1) The recent cloning of one of the SYM genes has greatly added our knowledge of the overall pathway of symbiotic signalling.

Two institutions simultaneously published papers describing the cloning of two orthologus receptor-like kinase genes essential in the microbial signal pathway, the nodulation receptor kinase (NORK) gene of M. sativa (Endre et al., 2002) and the symbiosis receptor-like kinase (SYMRK) gene of L. japonicus (Stacke et al., 2002). NORK and SYMRK are highly homologous, both are closely linked to the SHMT marker genes, and both have similar phenotypes, for these reasons NORK and SYMRK where are said to be orthologus. Analysis of SYMRK / NORK genes, in several Myc-/ Nod- mutants, confirmed that mutations in were responsible for the defective phenotypes. Mutations in SYMRK / NORK confer Myc-/ Nod- phenotypes defective in calcium spiking, membrane depolarisation, and extracellular alkalinisation placing then upstream of the earliest detectable response to symbiosis signals. To further clarify the position of the SYMRK gene within the symbiosis-signalling pathway the expression pattern of the symbiosis activated gene LB was observed, unlike wild type, SYMRK mutants didn't induce LB in response to NFs confirming SYMRK role in early signal transduction. The pea line p 55, mutated in PsSYM 19, has a similar phenotype the NORK / SYMRK mutant, PsSYM 19 is highly homologues to NORK / SYMRK at the protein level, and is also linked within 8 kb to the SHMT marker genes, these observation led to the conclusion that PsSYM 19 and NORK / SYMRK are orthologus genes (Stacke et al., 2002).

Similar studies have identified other possible NORK / SYMRK orthologus in M. truncatula (Mtdmi 12), M. alba, P. sativum, and Vicia hirsute. Hybridisation of the NORK gene with southern blots of total genomic DNA of several legumes produced bands in thirteen different genera of legume, strongly indicating the conserved presence of NORK homologues in nodulating plants (Endre et al., 2002). The protein structure of NORK / SYMRK has all the hallmarks of a receptor-like kinase, possessing a signal peptide, an extracellular domain, a transmembrane domain, and an intracellular protein kinase domain. Three extracellular leucine-rich repeats (L RRs) are found in the extracellular domain. The structure of NORK / SYMRK suggests a role in the detection and translocation of a ligand signal leading to calcium spiking.

However, several additional genes have been shown to be required for calcium spiking PsSYM 8 from pea (Walker et al., 2002), LjSYM 4 in L. japonicus, and MtDMI 1 in M. truncatula (Wais et al., 2000), mutations in all three of these genes confer a Myc-/Nod- phenotype. The orthologus relationship between these genes has not been established. It's possible that these additional SYM genes interact with NORK / SYMRK curtsey of their leucine-rich repeats although this has little evidence. The MtDMI 3 mutant is interesting in that it's phenotypically identical to the MtDMI 1 and MtDMI 2 mutants in all respects except calcium spiking (Catoira et al., 2000), MtDMI 3 mutants are able to induce calcium spiking in response to microbial signals.

If calcium spiking is a component in the direct signal transduction chain for NF signaling, then the position of DMI 3 would be downstream of calcium spiking, potentially consistent with a gene product involved in transduction of the calcium-spiking signal to downstream targets. Three other mutations have been reported which are phenotypically similar to MtDMI 3, they are PsSYM 9, PsSYM 30, and LjSym 30 (Walker et al., 2000). As a result all four of these genes are placed downstream of calcium spiking (fig 1). Further additions to the signalling pathway can be made studying Myc+/Nod- mutants, LjSYM 1, LjSYM 5, PsSYM 10, these mutants are able to form AM symbiosis but are defective in nodulation, failing to induce calcium spiking and root hair deformation in response to NF (Stougaard, 2001). This suggests these mutants are affected upstream of the SYM genes, and may be acting as NF binding proteins. Interestingly these mutations show no morphological deformation response to NFs, yet, as mentioned, mutations in the common SYM genes MtDMI 1, MtDMI 2, MtDMI 3, LjSYMRK, and LjSym 4 all show a root hair swelling response without root hair branching, suggesting that they are still able to partially detect and respond to NFs.

This indicates independent pathways for two aspects of NF induced root deformation, root hair branching being downstream of the SYM genes, and root hair swelling being upstream and independent of the SYM genes yet downstream of the hypothetical NF binding proteins (Fig 1). It's quite possible that these NF binding proteins can interact with NORK / SYMRK via the leucine rich region, to induce calcium spiking, and also interact with an a yet undiscovered signalling protein to induce root hair swelling. Fig 1. The common signalling pathway of leguminous plant genes during endosymbiotic infection. The bacteria nodulation factors (NFs) are shown entering the signal pathway at the extreme left, the mycorrhizal fungi signalling molecule is as yet unknown. Sym genes are shown in blue boxes, genes specific for NF recognition are shown in red boxes.

Mutations in NF binding protein genes, LjSYM 1, LjSYM 5, and PsSYM 10, confer a non-nodulating phenotype but retain the ability to form AM, they are defective in both NF induced root hair swelling and calcium spiking, these observations place these gene upstream of the SYM genes. Mutations in any of the SYM genes confer Myc-/Nod- phenotype. The NORK / SMIRK is shown in a dashed box, along with its proposed orthologs in pea and L. japonicus, mutations in these genes are shown to act upstream, along with LjSYM 4, MtDMI 1, and PsSYM 8, of calcium spiking, unlike LjSYM 30, MtDMI 3, PsSym 30, and PsSym 9, which are shown downstream. Root hair branching and root hair swelling are shown on different pathways as several mutants in the SYM genes respond with normal root hair swelling upon NF inoculation yet swelling is absent in NF binding protein mutants. The formation of infection threads cannot proceed without bacterial presence and therefore additional factors must play a role in this, these factors are marked as "?" . It was shown that a diffusible AM fungal factor induced MtENOD 11 expression, and this induction was independent of the common SYM, it's possible that this factor is the fungal equivalent to the bacterial Nod factors, and that ENOD induction branches away from the SYM cascade after a common fungal factor binding protein perception mechanism (shown as path a).

It's also possible that the AM fungal factor is perceived independently, and serves only to activate expression of the ENOD genes (shown as path b), SYM 8 has been shown to be essential for induction of ENOD 12 and ENOD 5 in both symbioses suggesting that it's interacting in the pathway leading to ENOD induction as well as the pathway leading to calcium spiking. This figure is adapted from Hirsch et al., 2001 It's also possible to obtain Nod- mutants which show wild type root deformation, for both swelling and branching, yet are unable to initiate infection thread or induce early nodulation genes. Transposon tagging of these mutant in L. japonicus identified the gene Lenin (Schauser et al., 1999), which encodes a transcription factor to induce ENOD genes in rhizobial infection. Homologues genes have been identified in pea, psSYM 7, and M. truncatula, MtNS P. Interestingly several of the same ENOD genes expressed downstream of NIN (ENOD 11, ENOD 40, and RIP 1) are also expressed in AM infection (Albrecht et al., 1998; Van Rhijn et al., 1997). Recent experiments using pMtENOD 11-gusA reporter gene proved that AM infection induces expression of pMtENOD 11-gusA even in Medicago truncatula lines possessing SYM mutations (MtDMI 1, MtDMI 2, MtDMI 3) (Kostu et al., 203). Furthermore, the it was shown that several membrane separations of the AM fungi and root hair were not sufficient to reduce this pMtENOD 11-gusA expression, indicating that the fungi produce a diffusible compound which is responsible for MtENOD 11 activation.

These observations imply that the pathway leading to MtENOD 11 induction, by the diffusible AM fungal factor, is separate from the Nod-dependant induction of MtENOD 11 in rhizobial symbiosis and separate from the common SYM cascade and calcium spiking (This independent path is shown on fig 1). It's possible that this diffusible fungal factor is the predicted AM signalling molecule, proposed to provide signalling from fungi to plant to activate the common SYM signal cascade in a similar way to NF, this suggests that the SYM cascade and MtENOD 11 expression pathways branch from a common AM fungal factor binding protein early in signal transduction (a in fig 1), it's also possible that the AM fungal factor acts solely on the pathway leading to MtENOD induction, whereas a different fungal factor may be the anticipated AM signalling molecule (b in fig 1). Further tests are required to distinguish between the two possibilities. However, it has also been shown that induction of PsENOD 12 and PsENOD 5 in both symbioses is absent in PsSYM 8 mutants (see above), this indicates sym 8 is common to both ENOD pathways and is added to fig 1. The discovery of the sym genes has important implications.

If one accepts that the mechanisms of AM and rhizobia infection converge, it implies that aspects of these mechanisms should be wide spread throughout plants that undergo either of the two symbioses. In keeping with this idea it was shown that rice were able to perceive the presence of NFs (Reddy et al., 1999). A Medicago ENOD 12 promoter-GUS complex was introduced into the non-legume rice plant. The addition of S. Meliloti NFs to the transgenic plant resulted in increased levels of GUS reporter gene expression, however the addition of chitooligosaccharide backbone molecules didn't induce ENOD 12-GUS expression. Taken at face value these finding suggest a mechanism by which rice can specifically perceive NFs and respond. Whilst it's obvious that a complete perception is not present, it's still an interesting finding and may have significance in attempts to create transgenic legumes capable of benefiting from nodulation.

It's also noted, that infection thread formation cannot occur in the absence of bacteria and therefore additional bacterial factors must come into play downstream of ENOD gene activation, the exact mechanism of this is not understood and is shown as "?" on fig 1. Gene hijacking in rhizobia evolution. The convergence of signalling between AM infection and rhizobial nodulation, as well as the fact that several early nodulin (ENOD) genes are expressed in both symbiosis, has led to the suggestion that rhizobia-plant symbiosis may have arisen from the much older AM-plant interaction. Several bacteria exist that posses the ability to reduce nitrogen, and if these bacteria were equipped with the ability to infect a plant root hair it's seems quite possible that this could lead to bacterial endosymbiosis. This of course leads to the question of how ancestral bacteria acquired the infection ability. Study of the nodA BC genes in Rhizobiaceae, genes encoding products in the pathway to NF production, suggests they are of external origin (Hirsch et al., 2001), their G + C content is below average, and codon usage is different from most chromosomal genes.

In addition, the chitin molecule product of the nod genes is quite unlike anything made by conventional bacteria, bacteria do not often posses chitin structures in their cell walls unlike fungi. NodC is an N-acetyl-glucosaminyl transferase that produces a chitin backbone, NodB removes an acetyl group from the terminal residue of the chitin oligomers, and NodA catalyses the transfer of a fatty acid chain onto the free amino group resulting from NodB. The action of NodA is very intriguing as there are few bacterial proteins known possessing such a mode of action, NodA like proteins have only been found in rhizobia and their source is a mystery. However, NodC protein is similar to several fungal chitin syntheses, suggesting rhizobia may have acquired these genes from an ancestral fungus. This idea partially explains the genetic overlap of the perception of these genes in legumes. It's possible that the diffusible factor discovered by Kostu and colleagues (Kostu et al., 2003) may represent the fungal signalling factor and further tests may prove or disprove this.

The identification of said factors may help improve our understanding of the overall evolution of nodulation. Where are we now and where are we going? It's clear that multiple genes are involved in both endosymbiotic interactions, and several of these genes overlap in function. The convergence of signalling mechanisms between the two-endosymbiotic symbioses offer prospect in the production of GM crop.

Utilisation of the rich nitrogen provided by rhizobial interaction would greatly reduce the amounts of fertiliser required in agricultural farming. Further study of the SYM genes in model organisms will help advance of understanding, but it's worth noting that model organism do not represent all species and further study on a species by species basis may be needed before agriculture can take advantage of bacterial symbiosis. The cloning of NORK / SYMRK has provided a surge of knowledge and interest into these species, further cloning of SYM genes should be equally as intriguing, and genome sequencing projects of L. japonicus and M. truncatula should facilitate this process. It's apparent that there is a considerable amount of cross talking between pathways an example is the possible presence of sym 8 in multiple pathways; this cross talk may cause constructing an overall model difficult. Arguments into the evolution of rhizobia are ongoing and whilst it's possible that these genes originated from fungal origin, it's also possible they stemmed from a bacterial source that has yet to be sequenced. Word Count: 2962 Albrecht, C. et al.

(1998) Endomycorrhizae and rhizobial Nod factors both require SYM 8 to induce the expression of the early nodulin genes PsENOD 5 and PsENOD 12 A. Plant J. 15,605-614 Brundrett, M. (2002) Coevolution of roots and mycorrhizal of land plants. New Phytol. 154,275-304 Cardenas, L. et al. (2000) Ion changes in legume root hairs responding to nod factors. Plant Physiol. 123,443-451 Catoira, R. et al.

(2000) Four genes of Medicago truncatula controlling components of a nod factor transduction pathway. Plant Cell 12, 1647-1665 Ehrhardt, D.W. et al. (1996) Calcium spiking in plant root hairs responding to Rhizobium nodulation signals. Cell 85,673-681 Endre, G., Kereszt, A., Kevel, Z., Mihacea, S., Kalo, P. and Kiss, G.B. (2002) A receptor kinase gene regulating symbiotic nodule development.

Nature. 417,962-966 Gianinazzi-Pearson, V. et al. (1998) Endomycorrhizae and rhizobial Nod factors both require SYM 8 to induce the expression of the early nodulin genes PsENOD 5 and PsENOD 12 A. Plant J. 15,605-614. Harrison, M. J (1997) The arbuscular mycorrhizal symbiosis: an underground association. Trends Plant Sci. 2, 54-60 Herendeen, P.S. et al.

(1999) A preliminary conspectus of the Allow flora from the Late Cretaceous (Late Santonin) of Central Georgia, U.S.A. Ann. MO Bot. Gard. 86,407-471 Hirsch, A.M. et al. (2001) What makes the rhizobia-legume symbiosis so special? Plant Physiol.

127, 1484-1492 Kosta, S. et al. (2003) A Diffusible Factor from Arbuscular Mycorrhizal Fungi induces Symbiosis-Specific MtENOD 11 Expression in roots of Medicago truncatula. Plant Physiol. 131,952-962 Nair, M.G., Safir, G.R. and Siqueira, J.O. (1991) Isolation amd identification of vesicular-arbuscular mycorrhiza stimulatory compounds from clover (Trifolium repens). Appl Environ Microbiol 57,434-439 No vero, M. et al.

(2002) Dual requirement of the LjSym 4 gene for mycorrhizal development in epidermal and cortical cells of Lotus japonicus roots. New Phytol. 154,741-749 Parniske, M. (2000) Intracellular accommodation of microbes by plants: a common developmental program for symbiosis and disease? Cur. Opin Plant Biol 3,320-328 Perret, X. et al. (2000) Molecular basis of symbiotic promiscuity.

Microbiol Mol Biol Rev. 64,180-201 Pueppke, S.G. and Broughton, W.J. (1999) Rhizobium sp. Strain NGR 234 and R. fred ii USDA 257 share exceptionally broad, nested host ranges. Mol Plant-Microbe Interact 12,293-318 Reddy, P.M. et al... (1998) Rhizobial lipochitooligosaccharide nodulation factors activate expression of the legume early nodulation gene ENOD 12 in rice. Plant J., 14,693-702. Remy, W. et al.

(1994) Four hundred-million-year-old vesicular arbuscular mycorrhizae. Proc. Natl. Acad.

Sci. U.S.A. 91, 11841-11843 Schneider, A. et al. (1999) Genetic mapping and functional analysis of a nodulation-defective mutant (sym 19) of pea (Pisum sativum L. ). Mol. Gen. Genet., 262, 1-11 Schauser, L., Rous sis, A., Stiller, J. and Stougaard. J. (1999) A plant regulator controlling development of symbiotic root nodules.

Nature 402,191-195 Smith, F.A. and Smith, S.E. (1997) Structural diversity in (vesicular) -arbuscular mycorrhizal symbioses. New Phytol., 137,373-388. Stougaard, J. (2001) Genetics and genomics of root symbiosis. Curr. Opin. Plant Biol.

4,328-335 Stracke, S. et al. (2002) A plant receptor-like kinase required for both fungal and bacterial symbiosis. Nature 417,959-962 Wais, R.J. et al. (2000) Genetic analysis of calcium spiking responses in nodulation mutants of Medicago truncatula. Proc. Natl.

Acad. Sci. U.S.A. 97, 13407-13412 Walker, S. A and Downie, J.A. (1999) Plant responses to nodulation factors. Curr Opin Plant Biol 2,483-9 Walker, S. A and Downie, J.A. (2000) Entry of Rhizobium leguminosarum bv. vicia into root hairs requires minimal Nod factor specificity, but subsequent infection thread growth requires nod or nodE. Mol Plant Microbe Interact 13,754-62 Walker, S.A., Vi prey, V and Downie, J.A. (2000) Dissection of nodulation signaling using pea mutants defective for calcium spiking induced by Nod factors and chitin oligomers.

Proc Natl Acad Sci USA 97, 13413-13418 Wegel, E., Schauser, L., Sandal, N., Stougaard, J. and Parniske, M. (1998). Mycorrhiza mutants of Lotus japonicus define genetically independent steps during symbiotic infection. Mol. Plant Microbe Interact.

11,933-936 Van Rhijn, P. et al. (1997) Expression of early nodulin genes in alfalfa mycorrhizae indicates that signal transduction pathways used in forming arbuscular mycorrhizae and Rhizobium-induced nodules may be conserved. Proc. Natl.

Acad. Sci. USA 94, 5467-5472.