Nodulin-24 follows a novel pathway for integration into the peribacteroid membrane in soybean root nodules.

Nodulin-24 is a nodule-specific protein of the peribacteroid membrane (PBM) in soybean. It has an apparent molecular mass of 33 kDa while its full-length cDNA encodes a polypeptide of only 24 kDa. In vitro transcription of nodulin-24 cDNA followed by translation resulted in a peptide translocated into microsomal membranes with cleavage of a signal sequence. The cleavage site of the signal sequence in nodulin-24 was determined to be between Ala (A25) and Arg (R26) by microsequencing of the [3H]leucine-labeled processed peptide. Fusion of the signal sequence of nodulin-24 with the beta-glucuronidase peptide prevented co-translational cleavage of the signal sequence although the translocation of the fused protein into microsomes occurred co-translationally. Trypsin treatment of membrane-translocated nodulin-24 did not result in any alteration in size suggesting that the newly synthesized peptide is fully protected in the membrane vesicle. Fusion of nodulin-24 with beta-glucuronidase also showed no change in size following trypsin treatment, suggesting that nodulin-24 has no membrane-spanning region. In addition, in vitro synthesized nodulin-24 was present in the supernatant fraction after sonication of microsomal membranes. Mature nodulin-24, on the other hand, is not solubilized from PBM by sodium carbonate (pH 11) or EGTA and is soluble only in detergent. These data suggest that nodulin-24 is synthesized as a lumenal protein in the endoplasmic reticulum and post-translationally attached to the membranes en route to the PBM. This processing results in a significant increase in the apparent molecular mass of nodulin-24 which may be due to the attachment of membrane lipids as this protein shares characteristics with membrane lipoproteins of many pathogenic bacteria.

Nodulin-24 Follows a Novel Pathway for Integration into the Peribacteroid Membrane in Soybean Root Nodules* (Received for publication, August 26, 1993, andin revised form, November 3, 1993) Choong-Ill Cheon, Zonglie Hong, and Desh Pal S. VermaS Nodulin-24 is a nodule-specific protein of the peribacteroid membrane (PBM) in soybean. It has an apparent molecular mass of 33 kDa while its full-length cDNA encodes a polypeptide of only 24 kDa. In vitro transcription of nodulin-24 cDNA followed by translation resulted in a peptide translocated into microsomal membranes with cleavage of a signal sequence. The cleavage site of the signal sequence in nodulin-24 was determined to be between Ala (A251 and Arg (R26) by microsequencing of the [SHlleucine-labeled processed peptide. Fusion of the signal sequence of nodulin-24 with the p-glucuronidase peptide prevented co-translational cleavage of the signal sequence although the translocation of the fused protein into microsomes occurred co-translationally. Trypsin treatment of membrane-translocated nodulin-24 did not result in any alteration in size suggesting that the newly synthesized peptide is fully protected in the membrane vesicle. Fusion of nodulin-24 with p-glucuronidase also showed no change in size following trypsin treatment, suggesting that nodulin-24 has no membrane-spanning region. In addition, in vitro synthesized nodulin-24 was present in the supernatant fraction after sonication of microsomal membranes.
Mature nodulin-24, on the other hand, is not solubilized from PBM by sodium carbonate (pH 11) or EGTA and is soluble only in detergent. These data suggest that nodulin-24 is synthesized as a lumenal protein in the endoplasmic reticulum and post-translationally attached to the membranes en route to the PBM. This processing results in a significant increase in the apparent molecular mass of nodulin-24 which may be due to the attachment of membrane lipids as this protein shares characteristics with membrane lipoproteins of many pathogenic bacteria.
Successful invasion of legume roots by rhizobia, followed by endocytosis of the bacteria into the host cytoplasm, results in the development of root nodules effective in nitrogen fxation (Verma, 1992;Brewin, 1991). Rhizobia are segregated inside the host cell into a subcellular compartment surrounded by a membrane, the peribacteroid membrane (PBM),' of host origin (Verma et al., 1978). The PBM plays critical roles in symbiosis as all metabolic exchanges between the two partners occur * This work was supported by Research Grant DCB 8819399 from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed: Ohio State Biotechnology Center, 1060 Carmack Rd., Columbus, OH 43210.
Many nodule-specific host proteins (nodulins) have been isolated (Delauney and Verma, 1988) and some of them, e.g. nodulin-24 and nodulin-26, have been localized to the PBM. Nodulin-26 is an intrinsic membrane protein and lacks a cleavable signal sequence (Miao et al., 1992). We have demonstrated that both amino and carboxyl ends of nodulin-26 face the host cell cytoplasm (Miao et al., 1992). On the other hand, nodulin-24 appears to be associated with the PBM and was suggested to be located on the surface facing the bacteroids . The nodulin-24 gene contains five exons, three of which (exons 2 4 encode a repeated amphipathic domain Katinakis and Verma, 1985). I n vitro translation of nodulin-24 mRNA in the presence of microsomal membranes suggested that nodulin-24 is processed co-translationally into a 20-kDa polypeptide . However, when PBM proteins were reacted with antibody against nodulin-24, the size of native nodulin-24 was found to be about 33 kDa, suggesting that this nodulin undergoes a significant post-translational modification . Nodulin-24 has homologies to nodulin-16 and a few glycine-rich plant proteins in signal peptide and COOH-terminal regions (Nirunsuksiri and Sengupta-Gopalan, 1990;Sandal et d . , 1992).
The nature of the PBM is unique as it possesses properties common to both plasma membrane and tonoplast (Verma et al., 1978;Mia0 et al., 1992). The biogenesis of the PBM compartment requires extensive vesicular transport (Cheon et al., 1993) and specific targeting of PBM and peribacteroid fluid nodulins. I t is not known how PBM nodulins are specifically targeted to this de nouo formed subcellular compartment. All PBM nodulins are synthesized on membrane-bound polysomes (Jacob et al., 1987) and co-translationally inserted into the membrane irrespective of whether they carry a n amino-terminal signal sequence or not. Co-translationally cleavable signal sequences of proteins contain a positively charged region and a hydrophobic core (von Heijne, 1983,19861, but the downstream domain flanking the cleavage site is not well characterized. The machinary to cleave the signal sequence is conserved between animals and plants (Chrispeels, 1991).
In this study, we examined synthesis, processing, and topology of nodulin-24 using i n vitro translated nodulin-24 peptides and isolated PBM. In addition, the cleavage site of the signal sequence in nodulin-24 was determined by microsequencing of [3H]leucine-labeled i n vitro translation product. The requirements for signal peptide cleavage were tested by protease-protection assay of nodulin-24 and p-glucuronidase fusion proteins. The results suggest that nodulin-24 is synthesized on the ER-bound ribosomes and is released into the lumen of the ER. Further post-translational processing, presumably in Golgi, attaches this protein to the membrane. Its overall hydrophobic and amphipathic character and the presence of a characteristic lipid-binding domain may allow this nodulin to be buried in the lipid of the PBM surface facing the bacteroids.
Plasmid Constructions-For in vitro transcription, a series of fusion constructs of nodulin-24 and P-glucuronidase were made as follows: a HindIII-BamHI restriction fragment from pN24, a pUCl9 derivative containing the nodulin-24 cDNA insert from pNod2O (Katinakis and Verma, 19851, was cloned downstream of the bacteriophage T7 promoter in pGEM2 (Promega Corp., Madison, WI). Constructs T1 and T2 were made by cloning cDNA fragments containing sequences corresponding to exon 1, and exons 1-2, respectively, of the nodulin-24 gene (see sequence accession no. MI05951 into XbaI and SrnaI-cut pBI221. The fragment containing exon 1 was obtained by digestion of pN24 with XbaI and DraI, followed by elution of a 0.1-kb band from an agarose gel, while the fragment containing exons 1-2 was obtained by digestion of pN24 with PstI and mungbean nuclease followed by digestion with XbaI. The entire length of the nodulin-24 cDNA, except the termination codon, was amplified by polymerase chain reaction and fused with the P-glucuronidase gene, resulting in the formation of construct T3. All the above constructs were subcloned into pGEM2 for in vitro transcription. In Vitro Danscription-Plasmids were linearized by cutting with appropriate restriction enzymes downstream of the gene of interest and transcribed with T7 RNA polymerase according to the manufacturer's instructions. In Vitro Dunslation and Processing ofthe Products-In vitro synthesized transcripts were translated in a rabbit reticulocyte lysate using either [3"Slmethionine or [3H11eucine and canine microsomal membranes (Promega Corp., Madison, WI). The protease protection assay was camed out as described (Spiess and Lodish, 1986). and the labeled peptides were immunoprecipitated (Anderson and Blobel, 1983) with nodulin-24 or P-glucuronidase antibody. f i r in vitro translation, the reaction mixture was centrifuged a t 356,000 x g (Beckman TL100) for 20 min a t 4 "C. The pellet was dissolved in STBS buffer (0.25 M sucrose, 10 nm Tris-HCI, 150 mM NaCI). Protease digestion with either L-Itosylamide-2-phenylethyl chloromethyl ketone treated trypsin or proteinase-K (final concentration of 100 pg/ml) was camed out on ice for 1 h in the absence or presence of Triton X-100 ('IX-100) at a final concentration of 1%. ConA-Sepharose 4B (Pharmacia LKB Biotechnology Inc.) binding was performed following immunoprecipitation protocol and replacing protein A with ConA-Sepharose. In vitro translated products were subjected to ten 5-s pulses of sonication (Sonifier 450, Branson Sonic Power Co., Danbury, CT) a t maximum output and then centrifuged at 356,000 x g for 30 min a t 4 "C. Samples were resolved in 10% or 14% SDS-polyacrylamide gels. Following electrophoresis, labeled products were analyzed by fluorography using EnHRance (DuPont-New England Nuclear).
Microsequencing-The [3Hlleucine-labeled nodulin-24, co-translationally processed in the presence of microsomal membranes, was resolved in a 10% SDS-polyacrylamide gel, and eluted by soaking in 0.1% SDS. The sample was adsorbed on a glass fiber filter that had been precoated with polybrene and subjected to Edman degradation on a protein sequenator (model 470A/900A, Applied Biosystems Inc., Foster City, CA). The radioactivity of fractions released in each cycle was measured by scintillation counting.
Western Blotting-PBM was prepared as described . The isolated PBM proteins were either extracted with 10 mM sodium carbonate, pH 11, 10 nm EGTA, or proteinase-K and subjected to sonication followed by incubation on ice for 15 min. The proteinase-K digestion was terminated by 1 m phenylmethylsulfonyl fluoride before adding SDS-sample buffer. Proteins were resolved on a 14% SDS-polyacrylamide gel, electroblotted to nitrocellulose membrane (Burnette, 1981), and detected using the ECGWestern blotting system (Amersham Sequence Analysis-The amino acid sequence of nodulin-24 was analyzed using Motifs program (Bairoch, 1991) of the GCG (Genetics Computer Group, Inc., Madison, WI) package. Corp.).

RESULTS
Cleavage of Signal Sequence a n d Topology of Nodulin-24-Nodulin-24 has been shown to be co-translationally processed into a 20-kDa polypeptide (Katinakis and Verma, 19851, but it is not known how it is associated with the membrane. We determined the topology of nodulin-24 in microsomal mem- lin-24 cDNA was transcribed in vitro using bacteriophage T7 polymerase, and the resulting transcript was translated in vitro using reticulocyte lysate and [:'sS]methionine in the absence or presence of microsomal membranes. The translation product was immunoprecipitated with antibody against a synthetic peptide corresponding to one of the repeated domains of nodulin-24. Translation products were assayed by tryptic digestion with and without the addition of "X-100. Mic, x-100. canine pancreatic microsomal membranes; D p , trypsin; Tx, Triton branes. In vitro transcribed nodulin-24 transcripts were translated using rabbit reticulocyte lysate and [35Slmethionine in the absence (Fig. 1, lane 1 ) or presence (Fig. 1, lane 2 ) of microsomal membranes, and immunoprecipitated with polyclonal antibody against nodulin-24 (see "Experimental Procedures"). As expected, nodulin-24 was processed into a 20-kDa polypeptide in the presence of microsomal membranes. Tryptic digestion of the translation mixture did not change the size of this polypeptide (Fig. 1, lane 31, although trypsin together with a nonionic detergent digested the processed nodulin-24 completely (Fig. 1, lane 4). This suggests that no part of the processed nodulin-24 protrudes from the ER into the cytoplasm.
In order to define the cleavage site of the signal sequence, the most probable cleavage site in nodulin-24 was first identified using a procedure described by von Heijne (von Heijne, 1986). Fig. 2A shows S(i) values a t each potential cleavage site which were obtained by summing the weights of residues including positions -13 and +2 relative to each site. The highest S(i) value assigned the cleavage site between Ala (A25) and Arg (R26). This site also satisfies the criteria of the (-3, -1) rule for cleavage (von Heijne, 1983;Perlman and Halvorson, 1983) with Val in the relative position -3 and Ala in the relative position -1. For experimental determination of the cleavage site, the [3H]leucine-labeled co-translationally processed product of nodulin-24 was subjected to Edman degradation. Radioactive peaks were found at the 3rd and the 21st cycles of sequential degradation which correspond to the positions of leucine residues in the deduced sequence of nodulin-24, if the cleavage occurred as predicted above. The data suggest that nodulin-24 is co-translationally processed into a mature polypeptide of 122 residues with arginine at the amino terminus.
The Entire Nodulin-24 Sequence Is Required for Signal Cleavage-The putative signal peptide of nodulin-24 identified above was tested for its role in targeting a protein to the ER. Different lengths of nodulin-24 cDNA were fused in-frame with the p-glucuronidase gene (Fig. 3A). Construct T1 contained the exon-1 sequence encoding the putative signal peptide plus 5 more residues fused with the p-glucuronidase gene; T2 had, in addition to the T1 sequence, the exon-2 sequence encoding 18 residues (repetitive domains in nodulin-24, Katanikis and  while T3 had the whole nodulin-24 coding region fused with p-glucuronidase. The in vitro translation product of P-glucuronidase mRNA alone could not be translocated into microsomal membranes as expected (Fig. 3B, panel d ) and served as a control. When T 1 was transcribed and translated in vitro, cleavage of the signal sequence was not observed. In-Mic--+ + + + + Sonic---1 +p -- stead, the size of the band increased due to co-translational glycosylation of p-glucuronidase (see Fig. 4; Iturriaga et al., 1989). The same results were obtained in the case of T2 which had 23 additional amino acids after the signal peptide. When a longer segment of nodulin-24 (containing exons -1, -2, and -3) was used, similar results were obtained (data not shown).
Translocation of the fusion products occurred in all cases as detergent addition to trypsin-protection assay showed complete digestion of the products. However, cleavage occurred only in the case of the T3 translation product. These results indicate that although the signal peptide is sufficient for translocation across microsomal membranes, the co-translational processing of nodulin-24 requires almost full-length of nodulin-24 flanking the cleavage site.
Nodulin-24 Is a Lumenal Protein in the ER and Is Modified Post-translationally to Become Membrane-associated-If nodulin-24 had any membrane-spanning region, construct T3 ( Fig.   3 A ) should have a stop transfer domain which would be detectable in the protease-protection assay. In that event, the entire P-glucuronidase protein should be located outside the membrane vesicles and thus become accessible to trypsin. However, no digestion of T3 product was observed (Fig. 3B). When microsomal membranes were solubilized by TX-100, the fusion protein was completely digested. Similar results were obtained using proteinase-K in place of trypsin (data not shown). These results suggest that nodulin-24 may be located inside the microsomal membrane or in the lumen of the ER. To test its location, we subjected microsomal membranes following in vitro translation, to sonication and centrifugation, and separated the microvesicles from the supernatant (Fig. 4, lams 3  and 4). Most of the nodulin-24 was found in the supernatant, indicating that nodulin-24 is not associated with the ER membranes but is present in the lumen. However, mature nodulin-24 isolated from root nodules is tightly attached to the membrane . Sodium carbonate (pH 11) (Fujiki et al., 1982) or EGTA treatment (van Renswoude and Kempf, 1984) of soybean PBM did not remove nodulin-24 from the membrane (Fig. 5), while TX-100 solubilized it along with the PBM (data not shown). These data suggest that nodulin-24 is a lumenal protein in the ER, but becomes modified post-translationally and attached to the membrane during its passage to the PBM. No co-translational glycosylation of nodulin-24 was detected using ConA (Fig. 4, lanes 5 and 6). Treatment of intact or sonicated PBM proteins with proteinase-K did not alter the size of nodulin-24 while nodulin-26 was cleaved (Fig. 6). This suggests that nodulin-24 is buried in the lipid layer and is not accessible to protease digestion. Addition of TX-100 to the reaction mixture resulted in complete digestion of nodulin-24 (data not shown), confirming that the protein is masked by lipids. Furthermore, nodulin-24 was found to have a homolo-

Attachment of Nodulin-24
gous domain to the lipid attachment site of prokaryotic membrane lipoproteins a t carboxyl-terminal region (Table I; Bairoch, 1991;Sankaran and Wu, 1993). These data indicate that nodulin-24 may be post-translationally anchored to lipid, possibly through this lipid attachment site, and become a membrane protein (see below) although no direct evidence is available for this phenomenon in eukaryotes.

DISCUSSION
Nodule development involves extensive membrane proliferation in the infected cells to enclose invading bacteria. Although the PBM is derived from the plasma membrane, progressive changes in composition of the PBM have been observed (Verma et al., 1978; including the acquisition of several new proteins such as nodulin-26 and nodulin-24. The expression of nodulin-26 and nodulin-24 is altered in ineffective nodules or nodules devoid of bacteria (Morrison and . If the vesicular transport of PBM proteins is retarding using antisense expression of rub7 gene, the late endosomes to Peribacteroid Membrane 6601 accumulate in the perinuclear region and these compartments become lytic, degrading endocytosed bacteria (Cheon et al., 1993). Nodulin-24 and nodulin-26 are targeted to the PBM via different mechanisms since nodulin-26 has no cleavable signal sequence (Miao et al., 1992) while nodulin-24 was shown to have a cleavable signal sequence. The amino acid sequences of nodulin-23 (Mauro et al., 1985) and nodulin-24 were analyzed using a weight-matrix approach (von Heijne, 19861, and potential cleavage sites were identified between amino acid residues Alaz0 and G1uZ1 for nodulin-23, and between Alaz5 and AI-^^ for nodulin-24. This suggests that nodulin-23 and nodulin-24 may be directed to the PBM by a similar sorting mechanism. We addressed the question of how nodulin-24 becomes attached to the PBM and increases its molecular mass from 20 to 33 kDa . Based on a hydropathy plot, the repeated domains at the amino-terminal region of processed nodulin-24 are highly amphipathic while the carboxyl half of this protein is hydrophobic. No membrane spanning region, which in most cases is a hydrophobic stretch of about 20 amino acid residues (Jennings, 1989), was observed in this protein . To test for the presence of a transmembrane domain using a protease-protection assay (Garoff, 19851, various regions of the nodulin-24 cDNA were fused with the p-glucuronidase reporter gene. When the entire cDNA of nodulin-24 was fused with the P-glucuronidase gene and the membrane-translocated product was treated with trypsin or proteinase-K (Fig. 3B ( d l ) , neither nodulin-24 nor P-glucuronidase was digested by the protease. This result suggested that either nodulin-24 has hydrophobic interactions with membrane lipids and is attached to the membrane or that it is a lumenal protein in the ER and is subsequently modified to become associated with the membrane. The experimental data suggest the latter possibility. Nodulin-24 was found in the supernatant fraction when the in vitro translation product, obtained in the presence of microsomal membranes, was sonicated and centrifuged (Fig. 4). These data suggest that nodulin-24 in the ER is not membrane bound but becomes attached to the membrane post-translationally along the sorting route to PBM. Proteinase-K treatment of PBM vesicles after sonication (Fig. 6) resulted in partial digestion of nodulin-26, while nodulin-24 was unaffected. Since nodulin-24 does not have any membrane spanning region, it may be covalently bound to and embedded in the membrane lipids which may be also responsible for the increase of its molecular mass. These lipids may also protect it from proteolytic digestion.
There are a few examples of lipids that act as membrane anchors. The variant surface glycoprotein of the parasitic protozoan, Tkypanosoma brucei, is linked to the membrane via the glycosyl-phosphatidylinositol (GPI) moiety (Low, 1989). Many

TARLE I
Presence of a consensus lipid attachment site in nodulin-24 first 6 residues can be any amino acid except D, E, R, or K. The following 2 residues can be any 2 amino acids of L, I, V, M, F, S, T, A, and G. The A prokaroytic membrane lipoprotein lipid attachment site was found in nodulin-24 sequence with no mismatch. In this motif (Bairoch, 1991) the 9th residue should be one of the following: I, V, M, S, T, A, G, or Q. The 10th position can be any ofA, G, or S. A cysteine residue in the last position of this motif is invariant and serves as the lipid attachment site.