THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 256, No. 6, Isue of

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 256, No. 6, I s u e of March 25, pp. 3092-3097, 1981 Printedin U.S.A.

Caulobacter crescentus Pilin PURIFICATION, CHEMICAL CHARACTERIZATION, AND NH2-TERMINAL AMINO ACID SEQUENCE OF A STRUCTURAL PROTEIN REGULATED DURING DEVELOPMENT*

(Received for publication, August 7, 1980, and in revised form, October 20, 1980)

John SmitS, Mark Hermodson& and Nina AgabianS From the *Department of Biochemistry SJ-70, University of Washington, Seattle, Washington 98195 and the §Department of Biochemistry, Purdue University, West Lafayette, Indiana 47907

The pili of Caulobacter crescentus are assembled dur- ing swarmer cell development at the differentiating cell pole. Under specific growth conditions, it was found that C. crescentus CB15 will produce insoluble pig- mented granules which entrap pili lost during cell growth along with other extracellular proteins. This provided a strategy for protein purification. Pilin was purified from these granules by gel filtration in the presence of high levels of detergent. A protein of the appropriate molecular weight for pilin was isolated by this procedure and demonstrated to be pilin by specific labeling of the intact pilus with ferritin-coupled anti- bodies. CB15 pilin has an apparent molecular weight of 8,000, is rich in hydrophobic amino acids (greater than 70%), and has a broad isoelectric focusing range cen- tered about a PI of 6.6. Periodic acid-Schiff reagent staining did not demonstrate carbohydrate modifica- tion of the monomer. The pilins of both CB15 and a related strain CB13 cross-react with anti-CB15 pilin antibody. Both strains also demonstrate similar RNA bacteriophage sensitivity, although there were signifi- cant differences in the amino acid composition, molec- ular weight, and other physical properties of the pilins from these two related strains. The NHz-terminal amino acid sequence of about 40% of the CB15 pilin molecule has been determined and bears some resemblance to that of some common pili; however, the sequences are not homologous and there is no indication of an unusual NHz-terminal amino acid in Caulobacter pilin.

Pili are long, slender filaments which are anchored in the membranes and protude from the surfaces of bacteria of many genera. In general, they are composed of a single protein species, pilin, which polymerizes to form the pilus structure. The structures themselves have varying characteristics and are usually distinguished by functional criteria. There are sex pili which play a role in the early events of cell conjugation and common pili which are often associated with attachment and agglutination properties of bacterial strains possessing these structures. Sex pili are generally 9 nm in diameter and have a monomer size of about M, = 10,000 to 12,000 (9, 14). There are generally only a few sex pili per cell (6). In compar- ison, common pili are generally about 6-7 nm in diameter and

* This work was supported by National Institutes of Health Grant GM 25527 and National Foundation-March of Dimes Grant 1-394 to N. A., National Institutes of Health Postdoctoral Fellowship AI 05946 to J. S., and National Institutes of Health Grant GM 24602 to M. H. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

several micrometers in length and are often found in large numbers on the cell. Their monomer protein ranges between M, = 17,000 and 19,000 (13, 18, 34).

Several pilins have been characterized by NH2-terminal amino acid sequence analysis, amino acid composition, and molecular weight determination. One class of common pili isolated from surprisingly diverse bacteria (Moraxella, Neis- seria, and Pseudomonas) yielded pilin of M , = 18,000, having highly homologous NH2-terminal amino acid sequences (12, 16, 27, 28). Each of these pilins had N-methylphenylalanine as the NHZ-terminal amino acid and from residues 6 through 23 contained only aliphatic hydrophobic amino acids. The amino acid compositions were similar in character, especially in high glycine and alanine contents, but not identical. Type 1 pilin isolated from a piliated, nonflagellated strain of Esch- erichia coli (B,,pf), on the other hand, had an NHZ-terminal amino acid sequence totally unrelated to the other sequences with no unusual amino acids and no long hydrophobic regions within the first 24 residues (16).

It has been postulated that at least some pili are able to extend and retract by mechanisms as yet to be determined (3, 4). If such a process does occur, the pilus must be able to pass through the cell membrane layers in a reversible manner. Thus, considerable interest focuses on the physical properties of this protein. One of the problems in studying the dynamics of pilus retraction, if it does occur, is that these apparent structural movements are not synchronous within a popula- tion of cells. In the case of Caulobacter crescentus, however, pili are assembled at the developing pole of swarmer cells, thus providing an interesting model for both the spatial and temporal regulation of pilus assembly. Furthermore, the ap- pearance of pili is coupled with the appearance of other differentiated structures at the developing cell pole. The abil- ity to assay pilin synthesis and expression would permit de- tailed analysis of the regulation of these various phenomena. However, in the past it has been difficult to isolate a sufficient quantity of pilin from Caulobacter to do most biochemical experiments. This is primarily due to the low number pro- duced per cell and the tendency for pili to adhere to many surfaces. Nevertheless, preliminary characterization of the pilin subunit of one C. crescentus strain, CB13, was possible (21).

In studying a related Caulobacter strain, CB15, we noticed that particles formed under specific growth conditions were able to entrap several extracellular proteins, one of which had a subunit molecular weight similar to that of the CB13 pilin.

We describe here a novel purification procedure based on this observation which produces milligram quantities of CB15 pilin in a rapid one-step purification. We also report a direct method to identify this protein as pilin using immunochem- istry and electron microscopy. We further describe the chem- ical characterization of this pilin, a partial sequence of the

3092

Caulobacter crescentus Pilin 3093

monomer, and a comparison with the pilin produced by a related strain, CB13.

MATERIALS AND METHODS

Bacterial Strains and Media-C. crescentus strain CB15 (ATCC 19089) was used in these studies for the production of the IRM.’ Cells were grown in a modified HIGG minimal medium (30), MsHIGG. MsHIGG contains 5 mM imidazole HCl (pH 7.0), 2 mM potassium

0.3% sodium glutamate (pH 7.0). For other studies HMG minimal phosphate, 1% modified Hutner’s mineral base (8), 0.3% glucose, and

(32) or a peptone yeast extract medium (1) was used. Strain CB13 and a derivative SW16-Pi1200 (21) were also used in some studies. Liquid cultures were incubated a t 30°C on a rotary shaker.

Production of Insoluble Red Material (ZRM)-A cycle of IRM production was as follows: 16 2-liter Erlenmeyer flasks containing 850 ml of M,HIGG medium were inoculated with 5 ml of an overnight CB15 culture. The flasks were shaken for 48 h, reaching an optical density (660 nm) of about 5.0. During the last 12 h of incubation, a red flocculent material was produced which formed loose granules of about 1 mm diameter. At the end of the growth cycle, cultures were poured through a Nytex Nylon mesh (Tetko, Inc.) which retained these granules. This material was washed on the mesh by irrigating with about 1 liter of H20, using a squirt bottle. The IRM was suspended in 50 rn l of H20 and centrifuged at 27,000 X g for 5 min. The pellet was washed twice in 50 ml of H 2 0 by centrifugation a t 750 X g and suspended in a small amount of H20. Sodium azide was added to 3 mM, and the slurry was refrigerated until used.

Gel Chromatography of IRM and Purification of Pilin-Eight ml of packed IRM, prepared from about 16 liters of cells as described above, were adjusted with concentrated solutions to yield a final volume of 12 ml containing 2% SDS, 0.1 M NaCl, 0.01 M disodium EDTA, and 0.01 M Tris-HC1 (pH 7.5). This mixture was shaken a t room temperature for 10 min, heated in a boiling water bath for 10 min, cooled to room temperature, and sonicated. The solution was then centrifuged a t 40,000 X g for 15 min to remove remaining insoluble material and applied directly to a Sephacryl 5-200 (Phar-

0.1 M NaCl, 0.01 M disodium EDTA, and 0.01 M Tris-HCI (pH 7.5). macia) gel filtration column (2.5 X 120 cm) equilibrated in 2% SDS,

The chromatography was done a t room temperature in the same buffer.

Fractions (5 ml) comprising the pilin peak (Fig. 1) were pooled and dialyzed against 14 liters of 3 mM NaN., for 16 h a t room temperature using Spectropor 3 dialysis membrane (Spectrum Medical Industries, nominal pore size 3,500 daltons). The pooled fractions were concen- trated by dialysis against dry Aquacide IIA (Calbiochem), redialyzed against eight changes of 14 liters of 3 mM NaNi3 over a period of 4 days, collected, and stored a t 4OC. These conditions resulted in a yield of about 15 mg of purified pilin.

Preparation of Antibodies to Pilin-Pilin, purified as described above, was used to immunize a New Zealand white female rabbit. Pilin (0.5 mg) in complete Freund‘s adjuvant was injected subcuta- neously. After 21 days a series of five sqbcutaneous injections of 0.25 mg of pilin in incomplete Freund’s adjuvant was given a t 7- to 9-day intervals. An adequate antibody response was detected from blood collected on days 51 and 63. The immunoglobulin fraction was par- tially purified from the serum by precipitation with 50% saturated ammonium sulfate followed by dialysis against NaCI/P,. For all ex- periments, the concentrations of all components in the fractionated sera were standardized to about twice the original serum concentra- tion. The double-immunodiffusion assay method of Ouchterlony (26). modified as previously described (20), was used to monitor the titer of the serum.

Preparation of Ferritin-Labeled Anti-pilin Antibody-Ferritin was coupled to anti-pilin antibody using the general procedure of Avremeas (2). A conjugation mixture contained 50 mg of ferritin (Polysciences) and 7 mg of ammonium sulfate-fractionated antiserum in 0.1 M potassium phosphate (pH 6.8) buffer containing 0.15 M NaCl in a total volume of 1.6 ml. A % o volume of 0.5% glutaraldehyde was added and the mixture was stirred at room temperature for 75 min. The preparation was dialyzed a t 4°C for 3 h against NaCI/P, contain- ing 0.2 M glycine, then overnight in NaCI/P, containing 3 mM sodium

’ The abbreviations used are: IRM, “insoluble red material” (de- scribed in text); SDS, sodium dodecyl sulfate; NaCI/P,, phosphate- buffered saline (10 mM potassium phosphate, pH 7.2, 0.15 M NaCI); BSA, bovine serum albumin; SDS-PAGE, polyacrylamide gel electro- phoresis in the presence of SDS.

azide, and again for 1-2 h against 0.05 M sodium phosphate (pH 7.5). The mixture was centrifuged a t 10.000 X g and the supernatant applied to a Sepharose 4B gel filtration column (1.6 X 34 cm) equili- brated with the phosphate buffer and chromatographed. Fractions between the excluded volume peak and the ferritin monomer peak were pooled and adjusted to 0.15 M NaC1, reduced in volume by dialysis against dry Ficoll 400 (Pharmacia), and dialyzed overnight against NaCl/P,. The preparation was adjusted to about 10 mg/ml of ferritin. assuming an extinction coefficient at 440 nm of 0.65 mg” ml- ’ (33), and stored at 4OC.

Ferritin-coniuEate Anti-pilin Antibody Labeling of Cells-Ap- proximately 50 p1 (about 5 2 10‘ cells) of cells were withdrawn from a vigorously growing culture and mixed with 150 pl of ferritin anti- pilin conjugate containing 1 mg/ml of BSA. This mixture was incu- bated without shaking for 15-30 min at room temperature. The cells and unreacted ferritin conjugates were then separated by Sepharose 2B gel filtration (column (0.7 X 24 cm)) in NaCI/P,. Fractions were monitored for the presence of cells by light microscopy and the cell peak was concentrated by a single centrifugation step (1 min at 80 v in a voltage-regulated Brinkman 5412 Microfuge). The cell pellet was suspended in a small amount of H 2 0 and negatively stained with ammonium molybdate. Preparations were analyzed using a Phillips EM 201 electron microscope.

Polyacrylamide Gel Electrophoresis-SDS-PAGE was performed according to published procedures (19, 35). Generally, gradients of 10-15% acrylamide were run. Proteins were stained with Coomassie brilliant blue (11) and periodic acid-Schiff reagent staining for car- bohydrate was done according to the method of Segrest and Jackson (31).

Isoelectric Focusing-Isoelectric focusing was performed using the method of O’Farrell (24). After focusing, the tube gels were fixed in 3.5% perchloric acid containing 30% methanol and heated at 65°C for 15 min. The gels were then stained according to the method of Otavsky and Drysdale (25). The pH gradient was calculated according to the method of O’Farrell (24).

Amino Acid Composition and Sequence Analysis-Purified pilin was hydrolyzed in 6 M HCl a t 110°C for 24 h and analyzed with a Durrum D500 amino acid analyzer. Amino acid sequence analyses were performed on a Beckman 890C Sequencer (15), and the products were identified by high pressure liquid chromatography (36). The quantity of the products was calculated a t each step and was more than 60% of the amount expected for the weight of the protein degraded in each of the early cycles of the degradation. Only one amino acid derivative was observed at each cycle in a yield compa- rable to the cycle before and after it. No positive identifications were made if the peak/background ratio was less than three. Tentative identifications were made where the data did not meet the above criteria, but where only one amino acid was observed.

“ -

RESULTS

Production a n d Isolation of Insoluble Red Material-The production of IRM is dependent upon the ammonium chloride concentration of the medium. In unmodified HIGG (30), the ammonium chloride concentration is 0.05% and IRM is incon- sistently produced. At ammonium chloride concentrations of 0.1%, IRM is never made, while removal of the ammonium chloride from the medium results in the consistent production of this material. Apparently, 0.05% ammonium chloride is a pivotal concentration for the production of IRM. In all cases, the medium also contains glutamic acid as a nitrogen source. Additionally, the rate of cell growth and aeration of cultures also appears to be important for optimal IRM production. Unmodified HIGG medium has 10 times less phosphate than M,HIGG, and while cultures will still achieve maximum den- sity, the growth rate decreases as phosphate becomes limiting and as the culture approaches stationary phase. Furthermore, if the culture volumes are too large, large clumps of IRM form which prevent efficient removal of intact cells during the washing procevs.

Composition of Insoluble Red Material-Granules of IRM are an intense red color, due to the presence of a pigment. The cellular origin of this pigment is not known; however, it is not proteinaceous and is virtually insoluble in water. IRM is also insoluble in ethanol, hexane, and acetone. I t is soluble

3094 Caulobacter crescentus Pilin

FIG. 1. Gel filtration of IRM using Sephacryl 5-200 and gel filtration buffer. The arrom indicates the position of pilin. It is the smallest protein present. The peaks to the right of pilin result from the pigment associated with IRM (see text). The distribution coeffi- cient K,. is the peak elution volume less the void volume divided by the total column volume less the void volume.

WAVELENGTKNM

FIG. 2. Absorption profile of the pigment associated with the IRM. It represents the peak seen by gel filtration at K., of 1.2 (see Fig. 1). The other species of this pigment show similar profiles, differing in the exact location of the major absorbance maxima and in small modulations in the 450 to 650 nm region (see text).

in 2% SDS, 6 M guanidine hydrochloride, 50% acetic acid, and dimethyl formamide. Gel filtration under conditions described under “Materials and Methods” indicates that its molecular weight is less than 2,000. As shown in Fig. 1, gel filtration resolved three to four pigmented peaks which were either red or green in color; the major absorption maximum varies with each peak, ranging from 401 to 405 nm (Fig. 2). The relative proportion of each species seems dependent upon pH; acidic pH results in a red color and basic pH a green color. At least one of the peaks is eluted beyond the included volume of the column (Fig. 1). These factors suggest that various interac- tions with the Sephacryl maxtrix are occurring and these peaks are probably not different molecular species being sep- arated according to molecular size.

The extreme insolubility of this pigment in water appeah to cause trapping of other particulates during growth to gen- erate filterable aggregates. Gel electrophoresis of IRM re- vealed a number of extracellular proteins (Fig. 3). The high molecular weight proteins form part of a regular surface array found in Caulobacter strain C B ~ ~ P . ~ The flagellins are also recovered in this material since cells discard their flagella during each growth cycle (20). Of particular importance for this study was the occurrence of a protein of the apparent molecular weight of pilin.

Constitutent proteins and pigment of IRM were separated by gel filtration in the presence of 2% SDS, 0.1 M NaCl, and

* J. Smit, D. A. Grano, R. M. Glaeser, and N. Agabian, manuscript in preparation.

0.01 M Tris. NaCl and Tris were necessary to achieve protein separation according to molecular weight in this maxtrix; when omitted, proteins ran as a single broad peak near the excluded volume. The apparent interaction of the red pigment with the Sephacryl filtration matrix was also important; frac- tionation using an agarose matrix of similar porosity (Sepha- rose 6B) resulted in a broad peak at the included volume which overlapped significantly with the pilin peak (data not shown). The material in the putative pilin peak indicated by the arrow in Fig. 1 was analyzed by SDS-polyacrylamide gel electrophoresis (Figs. 3 and 4). A single protein of apparent molecular weight of 8,000 was resolved in this peak fraction.

Identification of Pilin-Pili are not reliably assayed by electron microscopy and when IRM was examined by negative staining, intact pili were not detected. Therefore, other meth- ods were required to determine whether the protein in ques- tion was in fact pilin. The putative pilin, isolated by gel filtration as described above, was used to prepare specific antibody which was then coupled to ferritin as described under “Materials and Methods.” Fig. 5 demonstrates that ferritin-labeled anti-pilin antibody specifically decorates the pilus. No ferritin could be found attached to any other region of the cell. Control experiments revealed that neither the nonimmune serum nor anti-flagellin antibody labeled the pilus (data not shown).

A CB13 derivative,

a b c d e

SW16-Pil200 (21), was also examined

f g

92.5-

685-

41.0-

31 .O-

0

‘0

E 5 x

13.6-

8.5-

8 2 . 5 - 0 4

665- 4

41.0- .

-FLAGELLIN

1 a

-PILIN

13.6- e 4

8.5 -**l

FIG. 3 (left). SDS-PAGE of selected fractions of gel filtration of IRM using Sephacryl S-200 and the gel filtration buffer. The fractions are identified as an approximate K,, (refer to Fig. 1). a, molecular weight standards: phosphorylase b (92,500), BSA (66,500), alcohol dehydrogenase (41,000), DNase I (31,000), RNase A (13,600), and ubiquitin (8,500); 6,O.l; c, 0.2; d, 0.42; e, 0.8 1.2;g, unfractionated starting sample. Note that the pilin band is unusually wide. The lateral spreading often occurs during SDS-PAGE of this low molec- ular weight protein when no protein in that region is in adjacent channels.

FIG. 4 (right). SDS-PAGE of CB15 pilin isolated from IRM by gel filtration. The track containing pilin is flanked by molecular weight standards. The band at M , = 31,000 is caused by some spillover of the marker protein and is not a contaminant of the pilin prepara- tion.

Caulobacter crescentus Pilin 3095

55% hydrophobic amino acid content of CB13 pilin. Additional comparisons of the amino acid compositions of CB15 pilin

1 with CB13 pilin (Table I ) showed that a methionine residue is present in CB15 and absent in CB13 pilin. Additionally, proline and phenylalanine are absent in CB15 pilin, while CB13 pilin contains 2 residues of each. Both pilins are similar in that they contain low amounts of basic amino acids and are

FIG. 6. CB15 pilus labeled with anti-pilin antibody coupled ferritin. At numerous places along the length the pilus filament can be seen. Note the indications of a helical labelling pattern. Negativelv stained with ammonium molybdate. Bar indicates 0.25 pm.

FIG. 5. Labeling of C. crescentus CB15 cells with ferritin- coupled anti-pilin antibody. Cells were labeled as described under “Materials and Methods” and negatively stained with ammonium molybdate. Note the absence of ferritin on cell surface or the flagellum and the helical labeling pattern on the pilus (also see Fig. 6). Bar indicates 0.25 pm.

with ferritin-coupled anti-pilin antibody. Despite the fact that there are significant differences in the amino acid composition of pili from SW16 and CB15, the pili of SW16 were also heavily labeled with the antibody conjugate (data not shown). As shown in Figs. 5 and 6, ferritin-labeled antibody consist- ently produced a helical labeling of the pilus. The correlation of this result with the negatively stained appearance of unla- beled pili (Fig. 7) suggests that lateral aggregates of pilin can form a flat ribbon which is then twisted into a helical structure. The flat surfaces of the pilus are likely the most accessible region for antibody binding, resulting in the apparent helical appearance shown in Figs. 5 and 6. This labeling pattern was also seen for pili attached to cells (Fig. 5), so the laterally associated filaments may be an in vivo condition.

The pilus proved to be very fragile in these experiments. Usually cells are separated from unbound ferritin by several cycles of centrifugation. If treated in this fashion, no labeled pili remained associated with the cell. Presumably, the at- tached ferritin molecules increase the sensitivity of pili to shear forces. It is likely, in fact, that even the gentle procedures used in these studies resulted in considerable loss or breakage of these structures, since many swarmer cells were observed without pili.

Physical Characterization of Pilin-The apparent M , of 8,000 for CB15 pilin is in good agreement with the amino acid analysis which indicates that pilin contains 73 amino acids (Table I). It also compares well with the values of CB13 pilin (Table I) which has an apparent M , of 8,5OO and a correspond- ing 75 amino acids (21). Over 70% of the amino acid residues of CB15 pilin are hydrophobic amino acids with glycine and alanine predominating. This is significantly higher than the

‘4

FIG. 7. Electron micrograph of C. crescentus pili from cul- ture medium. Note the twisted ribbon-like appearance at left which separates into two filaments. This lateral association of filaments is commonly seen and may represent the structure of the organelle when attached to the cell (see text). A flagellar filament is included at left for size comparison. Negatively stained with ammonium mo- lybdate. Bar indicates 0.25 pm.

TABLE I Amino acid composition of C. crescentus pilins

Amino acid No. of residues

CRlS CH13”

Asx Thr Ser Clx Pro GlY Ala Val Met Ile Leu TYr Phe His LYS Arg

Total

3 3 4 3 0

16 20 8 1 4 4 3 0 1 0 3

73

4 10 6 7 2

12 13 5 0 3 6 1 2 1 2 1

75

“Taken from Ref. 21.

3096 Caulobacter crescentus Pilin

FIG. 8. Isoelectric focusing of purified CB15 pilin. The broad band between PI of 6.3 and 6.9 is seen but the faint “sub-bands” are very difficult to record photographically. Therefore, arrows are used to indicate their position at PI of 6.73, 6.65, 6.57, 6.50, 6.47, and 6.42.

5 SER GLU ARG GLY ALA MET VAL SER GLY ALA

10

1 5 ALA VAL ALA GLY GLY VAL GLY GLY ALA ASN

20

2 5 TYR ALA ARG ALA ALA ILE HIS GLY ALA (SER)

30

(X) GLY ALA FIG. 9. The NHr-terminal amino acid sequence of C. crescen-

t u CB15 pilin. Serine 30 was only tentatively identified and no amino acid was identified at cycle 31.

high in nonpolar amino acids; this similarity is also shared with type 1 and the F-pili of E. coli (5, 6).

A broad band between pH 6.3 and 6.9 was obtained by isoelectric focusing of purified pilin. Interspersed within the band were regions of greater protein density, forming at least 6 “sub-bands,” having reproducible PI values from experiment to experiment (Fig. 8). No bands were seen outside of this region and this apparent stuttering effect is not seen for other proteins in this system. The reason for this effect is unknown, but could reflect post-translational modification of the pilin, such as glycosylation. Periodic acid-Schiff reagent staining of pilin, however, did not reveal any carbohydrate, although many sugars commonly found in glycoproteins are not de- tected by this procedure. This result is also in contrast to the results obtained for CB13 pilin (21). Three sequencer degra- dations on pilin from different preparations gave consistent results. The sequence shown in Fig. 9 was obtained in excellent yield (near 100% of expected).

DISCUSSION

Analysis of the Caulobacter pilus has been hampered in the past by inability to easily obtain sufficient amounts of pilin for most studies. The pilus is expressed only during swarmer cell development and only a small number of pili are ever produced at the cell pole (21). This difficulty was overcome when we observed that if cultures were grown to very high density using unusual growth conditions, pigmented granules

which entrap pili are formed. IRM granules are readily sepa- rated from the cells and with no further treatment are pre- pared for gel filtration.

Fortuitously, there is no other extracellular protein of size similar to the pilin, permitting a single step purification of the pilus monomer protein. These circumstances now permit iso- lation of quantities of highly purified pilin limited only by the size of the culturing facilities and gel filtration apparatus.

The main drawback of this method is that only monomeric pilin is produced. We have been unable thus far to reconstitute the pilus using protein isolated in this fashion. Apparently this is not unusual; pilus reconsitution in other bacterial species has proven very difficult;’ and only a single report of success is known (5). Alternatively, it is possible that the protein does not accurately renature after removal of SDS, despite the fact that antisera raised against the purified pro- tein recognized the intact pilus.

The cellular location and function of the red pigment of IRM are not known. The sharp absorbance maximum at about 400 nm is not characteristic of any of the components involved with oxidative phosphorylation and electron transfer. I t is produced only as cultures approach stationary phase and as such may be a response to oxygen limitation, since Caulo- bacter is an obligate aerobe (29). Thus far, the pigment has only been elicited in HIGG medium. We are currently study- ing whether this is due to a component of this medium or is simply due to the fact that unusually high cell densities are produced in the medium. At this point, we are assigning this compound to the general class known as secondary metabo- lites, comparable to the pigments, alkaloids, and antibiotics produced by other bacterial species under certain conditions (IO, 22). The function and regulation of most of these com- pounds is poorly understood at the present time.

Thus far, attempts to purify pilin from strain CB13 using similar methods have not been successful. The red pigment in IRM which presumably binds sloughed off cellular compo- nents is either not produced by CB13 or is not produced using culture methods described. As a result, the extracellular ma- terial that this strain produces cannot be readily separated from the cells. Despite this difficulty, we are continuing our attempts to isolate a sufficient amount of CB13 pilin to permit sequence analysis.

The finding that ferritin-antibody labeling of pili occurred in a helical manner indicated a tendency for the pilus filaments to laterally aggregate. This would not be an unusual finding for unattached pili in the medium. However, pili attached to the cell also show this effect (Fig. 5) and show no indication of separate origins on the surface. I t may be that the pilus organelle is normally composed of two or more of these filaments. This fact could readily escape notice in many ex- periments because of the small diameter (about 4 nm) of the filaments.

The discovery that the antibody produced against CB15 pilin also cross-reacts with CB13 pilin is very interesting, since there are significant differences in the amino acid composi- tions of these two proteins and the proteins are very small. It is also known that both pilins are receptors for the RNA bacteriophage +Cb5, which infects both CB15 and CB13 (21). These findings favor the hypothesis that there exists a highly conserved region of homology between the two pilins recog- nized by the antibody, as well as a region of greater variability in primary structure.

A similar finding was seen in comparing the pilin proteins from Pseudomonas aeruginosa, Neisseria gonorrhea, and Moraxella nonliquifacieus (27). In that case the amino acid

‘I M. Achtman and W. Paranchych, personal communcation.

Caulobacter crescentus Pilin 3097

sequences for the three pilins were highly homologous for the fist 22 residues, while presumably considerable divergence occurs later in the sequence. Interestingly, the NH,-terminal sequences of these pilins were very hydrophobic, similar in character but not homologous to the CB15 pilin. In contrast, the remainder of the sequences for those strains had to be considerably more hydrophilic while in CB15 the highly hy- drophobic NH2-terminal sequence is no more hydrophobic than the remainder of this unusual protein, as deduced from amino acid composition.

As suggested for the other three pilins, this hydrophobicity may be related to the transport through the cell membranes and the assembly of the pilus on the outer membrane (27). If the Caulobacter pilus is capable of retraction, as hypothesized for other pili (3, 4), it is reasonable to expect a hydrophobic protein, which might allow storage of the monomer in the membrane system.

The CB15 pilin is distinguished from the three pilins dis- cussed above by the absence of an N-methylphenylalanine at the NH, terminus. The high yields obtained in sequence analysis argue that degradation of the NH, terminus did not occur during the long process of granule production, gel fdtra- tion, and extensive dialysis; however, this possibility cannot be completely ruled out. I t is notable however, that the E. coli pilin also does not contain this unusual amino acid (16).

The NH2-terminal sequence of CB15 pilin shows no homol- ogy to any of the other pilin sequences that have been deter- mined to date. Only the general physical characteristics of CB15 pilin are comparable to the other pilins studied. I t appears that there may be no simple way to correlate structure and function among bacterial pili as was initially suspected, based on the high degree of homology of some of the NH2- terminal sequences. We also note that a preliminary analysis of secondary structure by the predictive methods of Chou and Fasman (7) did not reveal any extensive regions of a-helical or P-pleated sheet structure, largely due to the high percent- age of glycine residues. I t is clear that further studies on a still wider variety of pili will be necessary to deduce functional aspects from sequence information.

The function of the Caulobacter pilus remains unknown. The physical properties of the monomer do not clearly favor analogy to either common or sex pili of other organisms. Since exchange of genetic information has been reported for Cau- lobacter (17, 23), we are currently examining the possibility that the pilus has a role in mediating this process. Finally, since the pilus is involved in the differentiation process of Caulobacter, there may be unique functions for this organelle that have no analogy in other bacterial systems. The availa- bility of purified pilin and antibody has permitted studies directed to understanding the role of the pilus in Caulobacter development. A study has been made of the periodicity of synthesis of pilin throughout the cell cycle and the relation- ship to other developmental proteins and will be reported e l~ewhere .~

J. Smit and N. Agabian, manuscript in preparation.

Acknowledgments-We thank Richard Granberg for performing the amino acid analysis.

REFERENCES

1. Agabian-Keshishian, N., and Shapiro, L. (1971) Virology 44, 46-

2. Avrameas, S. (1969) Immunochemistry 6, 43-52 3. Bradley, D . E. (1972) J. Gen. Microbiol. 72, 303-319 4. Bradley, D. E. (1972). Genet. Res. 19, 39-51 5. Brinton, C. C., Jr. (1965) Trans. N. Y. Acad. Sci. 27, 1003-1054 6. Brinton, C. C., Jr. (1971). CRC Crit. Reu. Microbiol. 1, 105-160 7. Chou, P. Y., and Fasman, G. P. (1978) Adu. Enzymol. 47, 45-148 8. Cohen-Bazire, G., Sistrom, W. R., and Stanier, R. Y. (1957) J.

9. Date, T., Inuzuka, M., and Tomoeda, M. (1977) Biochemistry 16,

10. Drew, S. W., and Demain, A. L. (1977) Annu. Rev. Microbiol31,

11. Fairbanks, G., Steck, T. L., and Wallach, D. F. H. (1971) Bio-

12. Frgholm, L. T., and Sletten, K. (1977) FEBS Lett. 73, 29-32 13. Frost, L. S., and Paranchych, W. (1977) J. Bacterid. 131, 259-

14. Helmuth, R., and Achtman, M. (1978) Proc. Natl. Acad. Sci.

15. Hermodson, M., Schmer, G., and Kurachi, K. (1977) J . Biol.

16. Hermodson, M. A., Chen, K. C. S., and Buchanan, T. (1978)

17. Jollick, J. D., and Tran, T. Q. (1975) J . Gen. Microbiol. 91, 183-

18. Kumazawa, N., and Yanagawa, R. (1972) Infect. Zmmun. 5,27-30 19. Laemmli, U. K. (1970) Nature 227,680-685 20. Lagenaur, C., and Agabian N . (1976) J. Bacteriol. 128,435-444 21. Lagenaur, C., and Agabian, N. (1977) J. Bacteriol. 131, 340-346 22. Malik, S. V. (1980) Trends Biochem. Sci. 15,68-72 23. Newton, A., and Allebach, E. (1975) Genetics 80, 1-11 24. O’Farrell, P. H..(1975) J. Biol. Chem. 250,4007-4021 25. Otavsky, W. I., and Drysdale, J. W. (1975) Anal. Biochem. 65,

533-536 26. Ouchterlony, 0. (1949) Acta Pathol. Microbiol. Scand. 26, 507-

515 27. Paranchych, W., Frost, L. S., and Carpenter, M. (1978) J. Bacte-

riol. 134, 1179-1180 28. Paranchych, W., Sastry, P. A., Frost, L. S., Carpenter, M., Arm-

strong, G. D., and Watts, T. H. (1979) Can. J . Microbiol. 25, 1175-1181

53

Cell Comp. Physiol. 49, 25-68

5579-5585

343-356

chemistry 10,2606-2617

269

U.S.A. 75, 1237-1241

Chem. 252,6276-6279

Biochemistry 17,442-445

187

29. Poindexter, J. S. (1964) Bacteriol. Reu. 28, 231-295 30. Poindexter, J. S. (1978) J . Bacteriol. 135, 1141-1145 31. Segrest, J . P., and Jackson, R. L. (1972) Methods Enzymol. 28,

54-63 32. Shapiro, L., Agabian-Keshishian, N., Hirsch, A,, and Rosen, 0.

M. (1972) Proc. Natl. Acad. Sci. U S A . 69, 1225-1229 33. Singer, S. J., and Shick, A. F. (1961) J. Biophys. Biochem. Cytol.

9, 519-530 34. Swaney, L. M., Liu, Y.-P., To, C.-M., To, C.-C., Ippen-Ihler, K.,

and Brinton, C. C., Jr. (1977) J. Bacteriol. 130, 495-505 35. West, D. C., Lagenaur, C., and Agabian, N. (1976) J . Virol. 17,

568-575 36. Zimmerman, C. L., Apella, E., and Pisano, J. J. (1977) Anal.

Biochem. 77,569-573