Preliminary analysis of polyhydroxyalkanoate inclusionsusing atomic force microscopy

Douglas Dennis a;�, Caroline Liebig b, Tara Holley b, Kara S. Thomas b,Amit Khosla b, Douglas Wilson b, Brian Augustine c

a Department of Life Sciences, Arizona State University West, P.O. Box 37100, Phoenix, AZ 85069-7100, USAb Department of Biology, James Madison University, Harrisonburg, VA, USAc Department of Chemistry, James Madison University, Harrisonburg, VA, USA

Received 5 May 2003; received in revised form 8 July 2003; accepted 15 July 2003

First published online 26 August 2003

Abstract

Atomic force microscopy analysis of polyhydroxyalkanoate (PHA) inclusions isolated from sonicated Ralstonia eutropha cells revealedthat they exhibit two types of surface structure and shape; rough and ovoid, or smooth and spherical. Smooth inclusions possessed linearsurface structures that were in parallel arrays with 7-nm spacing. Occasionally, cracks or fissures could be seen on the surface of the roughinclusions, which allowed a measurement of approximately 4 nm for the thickness of the boundary layer. When the rough inclusions wereimaged at higher resolution, globular structures, 35 nm in diameter, having a central pore could be seen. These globular structures wereconnected by a network of 4-nm-wide linear structures. When the inclusions were treated with sodium lauryl sulfate, the boundary layerof the inclusion deteriorated in a manner that would be consistent with a lipid envelope. When the boundary layer was largely gone, 35-nm globular disks could be imaged laying on the surface of the filter beside the inclusions. These data have facilitated the development ofa preliminary model for PHA inclusion structure that is more advanced than previous models.7 2003 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved.

Keywords: Atomic force microscopy; Poly-3-hydroxyalkanoate; Inclusion

1. Introduction

Bacterial polyhydroxyalkanoates (PHAs) are biodegrad-able polyesters that are found as intracellular inclusionsthat are generally between 200 and 500 nm in diameter[1,2]. At present, very little is known about their structureand biogenesis. In Ralstonia eutropha, three proteins havebeen shown to exist at their surface, PhaC, PhaP andPhaR. PhaC is the PHA synthase and is responsible forthe formation of the polyester from 3-hydroxy fatty acyl-CoA molecules [2,3]. PhaP is thought to be a structuralprotein because it is found in very large quantities inPHA-accumulating cells and has profound e¡ects on theshape and number of PHA inclusions when it is deleted or

overexpressed [4]. PhaR regulates the expression of PhaP[5,6]. Early electron microscopic studies (carbon replica)on inclusions from Bacillus megaterium supported the ex-istence of a membrane layer at the surface of the inclusion[7], but later studies on inclusions from Pseudomonad spe-cies imaged a regular lattice-like surface architecture thatis reminiscent of a bacterial S-layer [8^12]. Because of this,some researchers suggested that a protein lattice covers theinclusions, most logically comprised of PhaP. The possi-bility of whether the boundary layer is either membrane orprotein has been somewhat clari¢ed in that researchershave determined that boundary layer thickness is 4 nm[13,14], a fact that would seem to exclude the possibilityof a lipid bilayer surrounding the inclusion because lipidbilayers are approximately 8 nm in thickness. Mayer andHoppert have o¡ered strong theoretical arguments basedon hydrophobic interactions of membranes as to why thisboundary layer is most probably a lipid monolayer. None-theless, de¢nitive evidence on the composition of theboundary layer and the arrangement of proteins at its sur-

0378-1097 / 03 / $22.00 7 2003 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved.doi :10.1016/S0378-1097(03)00610-4

* Corresponding author. Tel. : +1 (602) 543 6934;Fax: +1 (602) 543 6073.

E-mail address: douglas.dennis@asu.edu (D. Dennis).

FEMSLE 11172 4-9-03

FEMS Microbiology Letters 226 (2003) 113^119

www.fems-microbiology.org

face is scarce, and several models of inclusion structurecontinue to exist [2,8,9,11,12].Recently, atomic force microscopy (AFM) studies on

inclusions from Comomonas acidovorans have given sup-port to the existence of an envelope around the inclusion[15]. The possible existence of a lipid boundary layer isquite interesting because a corollary implication is thatthere would have to be structures that would allow forthe ingress of substrates for polymerization, as well asegress of depolymerized monomers to be released intothe cytoplasm. The recent AFM work also noted the pres-ence of uncharacterized 30-nm ‘globular’ structures on thesurface of the inclusions.The use of AFM to analyze PHA inclusions surfaces is

particularly appropriate because of the minimal amount ofpreparation necessary to image the inclusions. Past inclu-sion models su¡er from the fact that they were developedfrom electron micrographs of PHA inclusions that re-quired signi¢cant sample preparation and manipulationprior to microscopy, possibly disrupting the integrity ofthe inclusion surface. In this study, we present AFM im-ages that con¢rm the existence of the boundary layer andwhich reveal a network of organized structures on thesurface of the inclusion that may function in structureand metabolism, as well as facilitating movement of mol-ecules through the boundary layer.

2. Materials and methods

2.1. Bacterial strains and culture conditions

R. eutropha was grown in Luria broth containing noNaCl (LBN) for routine growth and was maintained onLBN agar. For induction of PHA accumulation the bac-teria was grown in Schlegel’s nitrogen-limited minimal me-dia containing 0.1% (w/v) ammonium sulfate and 0.5%fructose [16].

2.2. Inclusion preparation

R. eutropha cultures were harvested when inclusionswere clearly visible intracellularly (light microscopy),which was usually about 24 h post-inoculation when theSchlegel’s nitrogen-limited media was used. Five microli-ters of culture was added to 5 ml of 50 mM Tris (pH 8.0)^1 mM EDTA (or 50 mM Tris (pH 8.0)^0.15 M NaCl) andthe suspension was added to a Falcon 2059 polypropylenetube and sonicated using an Tekmar 400 watt sonicatorwith a mid-sized probe. Sonication was done in an ice bathfor 5 min using 5-s-on and 7-s-o¡ cycles. To rid the soni-cated suspension of intact cells, the tube was centrifuged at1000Ug for 10 min at 4‡C. The top 4 ml of the supernatewas removed to a new tube and placed on ice. For AFMstudies, varying amounts of this suspension were added to10 ml of 50 mM Tris (pH 8.0)^1 mM EDTA and this was

¢ltered through a 250-nm pore size polycarbonate ¢lter.After the initial 10 ml, the ¢lter was washed with 2U5 mlof TE, followed by 5 ml of water. The ¢lter was immedi-ately removed, adhered to a metal puck (12 mm; TedPella, Inc) and was directly imaged by AFM.

2.3. Sodium dodecyl sulfate (SDS)-mediated disintegrationof boundary layer

This procedure was the same as above, except that the4 ml of supernate removed after the 1000Ug centrifuga-tion was now added to 36 ml of a prewarmed (45‡C) SDSsolution such that the ¢nal SDS concentration was 0.5%(w/v). The suspension was gently mixed and incubated at45‡C for 5 min. Varying amounts of the suspension wereremoved and ¢ltered through 250-nm pore size polycar-bonate ¢lters, followed by two 5-ml washes with deionizedwater. Filters were processed as above.

2.4. AFM

AFM was performed on a Digital Instruments Multi-mode instrument operating in tapping mode to acquireboth height and phase images simultaneously. All imagingwas performed in air at room temperature using silicontapping mode probes manufactured by Olympus and pur-chased from Digital Instruments. Typical resonance fre-quencies for the probes were V250^300 kHz. Phase imag-ing was obtained by applying a relatively large tappingforce (low setpoint) to the surface of the inclusion bodies.Typical scan speeds were 0.3^0.7 Hz.

3. Results

In general, two types of inclusions (Fig. 1A,B.) wereobserved by AFM. The majority of inclusions tended tobe ovoid in appearance and have a ‘rough’ or ‘bumpy’surface. Occasionally, another type of inclusion that isrounder and had a smoother surface was seen. The‘smooth’ inclusions had a di¡erent pro¢le in that theyare more ‘mushroom-like’ and did not extend as highabove the surface of the ¢lter (cross-section data notshown). Generally, their height pro¢le was about half ofthe pro¢le of rough inclusions. The proportion of smoothinclusions in the preparation increased if, 1) bacterial cellsstored at 4‡C for several days were used, 2) the cells weresubjected to longer periods of sonication, or 3) higherlevels of EDTA were used (5 or 10 mM). When the ¢lterscontaining inclusions were stored at room temperature fora week and re-imaged, wide ¢ssures or cracks on thesurface of the inclusion could be seen, presumably dueto drying. The surface of the sublayer at the bottomof these ¢ssures is consistently 4 nm below the surface ofthe ‘rough’ inclusion (cross-section analysis, data nowshown).

FEMSLE 11172 4-9-03

D. Dennis et al. / FEMS Microbiology Letters 226 (2003) 113^119114

3.1. High resolution studies

When the ‘rough’ and ‘smooth’ inclusions were imagedusing AFM at resolutions near the limit of the instrument,a di¡erent type of structure could be identi¢ed on thesurface of the inclusion. The smooth inclusions had linearstrands in parallel arrays traversing the surface of the in-clusions (Fig. 1C). These parallel arrays did not extenduninterrupted across the surface of the inclusions, butcould be seen traversing the surface of the inclusion inregions that were then interrupted by another set of lineararrays traversing a di¡erent direction. It would be analo-gous to the strokes of a paintbrush across the surface ofthe inclusion. The distance between the lines of the parallel

array was quite consistent and was measured by cross-sec-tional analysis at 7 nm (data not shown). The ‘rough’inclusions exhibit quite a di¡erent surface morphology.The surfaces of these inclusions were populated with glob-ular structures that were roughly spherical and which hada central pore (Fig. 1D). The globular structures wereconsistent in size, approximately 35 nm in diameter.Cross-sectional analyses of these structures revealed thatthe average pore diameter of the structure was about 15nm (at the surface). The collar of the structure was about12^15 nm in diameter and extended 4^12 nm above thesurface of the inclusion (data not shown). Generally, manypores can be seen and they appear to be connected bylinear structures that traverse the surface of the inclusion,

Fig. 1. AFM images of PHA inclusions. AFM height (A) and phase (B) images of two types of PHA inclusions isolated from sonicated cells. In theheight image the brightness of the image is directly related to the height of the inclusion. Terracing at the edge of the inclusion in the phase image isan artifact and should be disregarded. C: AFM high resolution phase image of the surface of a smooth inclusion showing 7-nm parallel arrays.D: AFM phase image of rough inclusion showing 35-nm globular structures with central pore.

FEMSLE 11172 4-9-03

D. Dennis et al. / FEMS Microbiology Letters 226 (2003) 113^119 115

making it appear that there is a cytoskeletal structure en-compassing the inclusion (Fig. 2). The average width ofthese structures is about 4 nm, but they can be wider,particularly at the points that they connect to the ‘porin-

like’ structures. In the highest resolution images, the sameparallel linear arrays (7 nm spacing) seen on smooth in-clusions could be imaged beneath the cytoskeletal-like net-work (Fig. 2).

Fig. 2. High resolution AFM phase image of PHA inclusion. Speci¢c structures mentioned in text: 1: Parallel arrays with 7-nm spacing; 2: linear net-work-like structures ; 3: porin-like structures with central pore. Network structures appear to be more prominent at the edge of scan because of tip con-volution.

Fig. 3. AFM phase images of inclusions treated with warmed 0.5% SDS. The left image shows the boundary layer in earlier stages of deterioration,where the right image shows an inclusion in the later stage of deterioration. Cross-sectional analysis of these images reveals the thickness of these layersto be approximately 4 nm.

FEMSLE 11172 4-9-03

D. Dennis et al. / FEMS Microbiology Letters 226 (2003) 113^119116

3.2. SDS studies

Because our preliminary data suggested that the ‘rough’inclusion possessed a boundary layer that is 4 nm thick(data from dried inclusions); and because this thicknessagrees well with prior studies on the inclusion boundarylayer[13,14], we pursued studies aimed at disruption of thisboundary layer, and subsequent AFM imaging of this dis-ruption. Inclusions treated with 0.5% (w/v) SDS at 45‡Cfor 5 min displayed a consistent deterioration process.First small holes were opened in the boundary layer, gen-erally around the globular structures (data not shown). Atthis time, the parallel array could often be observed eventhough it was still underneath the boundary layer. As thedeterioration progressed the boundary layer was made tolook like a layer of Swiss cheese in that circular areas ofdeterioration could be imaged (Fig. 3). The holes gradu-ally enlarged until there are only stringy strands of bound-ary layer left clinging to the surface of the inclusion.Strands of material were frequently seen connecting adja-cent inclusions. Though the globular structures could beimaged early in the deterioration, they could not be im-aged very late in the deterioration process. In images thatcontain inclusions that have largely lost their boundarylayer, globular discs were observed scattered over the sur-face of the polycarbonate ¢lter near the inclusions (Fig. 4),but not in areas of the ¢lter that did not have inclusions.These structures were approximately the same diameteras the 35-nm globular structures imaged on intact inclu-sions.

4. Discussion

The AFM data generated in this research project pro-vides a di¡erent analysis of PHA inclusions from priorstudies using electron microscopy. A large measure ofthis is due to the fact that the inclusions can be har-vested and imaged with minimal preparation, therebyleaving structures intact that may be removed by therelatively harsh preparative procedures required in elec-tron microscopy. Since AFM is a relatively new technol-ogy, there are issues that must be considered when ap-plying the technology to a new system. For instance,AFM is a probe-limited technique that su¡ers from tipconvolution e¡ects in which the geometry of the probecan distort the horizontal measurements made at the sur-face of an object [17^19]. This can be circumventedsomewhat by using the distance between maximumheight features to measure horizontal distances, butmany horizontal distances must be viewed as estimates,rather than exact measurements. For convex samplesthere is an ‘edge’ e¡ect in which the probe does notmaintain a consistent contact with the surface as it tra-verses down the side of the object and the angle of con-tact becomes increasingly non-perpendicular. For thisreason, only the data gathered from the relatively £atcenter of the object was included in the analyses de-scribed in this report. Most of the images displayed inthis report are the result of phase imaging. Phase imag-ing is a complementary scanning probe technique thatcan be correlated to di¡erences in hardness or elasticityof an object and has been widely used for polymericmaterials [20]. The technology is subject to artifactualimages and must be carefully compared with height im-ages in order to be considered valid. Though phase im-ages are largely used in this report (because of theirclarity), each one was carefully compared against a rep-resentative height image before it was deemed to bevalid.Our preliminary data allows us to build a somewhat

speculative, but testable, model for inclusion structureand function. First, we suggest (based on our and others’data) that it is proven at this point that there is a 3^4-nm-thick boundary layer that surrounds every inclusion. Mostlikely the ‘rough’ inclusions imaged by AFM in our ex-periments have a boundary layer, whereas the ‘smooth’inclusions have lost it. The chemical nature of this bound-ary layer is not known, but good arguments based onstructure and the hydrophobicity of the interior of theinclusion have indicated that it is a lipid monolayer [13].Given this, the next question that must be asked is howsubstrates gain access, and depolymerized products gainegress though this boundary layer. Our AFM imagesclearly show structures, reminiscent of porins (but withlarger dimensions than known porins), which are possiblecenters for access/egress. This possibility is supported bythe fact that we, and others, have isolated porin-like mol-

Fig. 4. AFM phase image of SDS-treated inclusions showing 35-nmspherical structures on the surface of the ¢lter adjacent to the inclu-sions.

FEMSLE 11172 4-9-03

D. Dennis et al. / FEMS Microbiology Letters 226 (2003) 113^119 117

ecules from the surface of inclusions [21,22]. The porin-like structures appear frequently and randomly on the in-clusion and are connected by linear spokes that are about4 nm in width. Beneath these structures can be seen par-allel arrays that have a 7-nm spacing. It seems likely thatthe parallel arrays are the polymer itself that is being im-aged. PHAs form a well-known lamellar spherulitic micro-structure [23^25] and we have observed a structure similarto these parallel arrays in spun-cast thin ¢lms of PHB-co-V [26]. We further suggest that globular structures tra-versing the surface of the parallel arrays are the proteinmachinery that mediates structure, synthesis, and depoly-merization. As such, a large portion of this is likely to becomposed of PhaP. The estimated width of PhaP, basedon its molecular mass, agrees with the width of the linearstructures (4 nm). These structures must not be verytightly bound to the polymer because inclusions thathave lost their boundary layer also lose this network.One explanation for this phenomenon would be that thestructures are linked to the boundary layer. The ease withwhich these structures are lost o¡ the inclusion surfacewould explain why they have not been imaged by electronmicroscopy with its inherently rigorous preparation proce-dure. Based on the commonality of membrane-spanningfunctional tunnel complexes (such as ATPase) in biology,we suggest that the spherical structures containing thecentral pore are synthesis/depolymerization centers. There-fore, in this model as the inclusion grows new proteinnetworks are laid across the surface and these synthesis/biosynthesis centers are added to these tracks by a processthat is unknown. As sites of synthesis and degradation itwould be expected that PHA synthetic and/or degradativeenzymes are transiently or permanently associated withthis complex. The globular structures that can be seenlittering the surface of the ¢lter after SDS treatment areintriguing, but problematic. Other than the fact that theyare of the same diameter as the porin-like structures seenon the surface of the inclusion, there is no data to suggestthat this is their origin. However, we are experimentallypursuing these as possible protein complexes for synthesisand depolymerization, which are held in place by a proteinsca¡olding system. This model agrees with the Mayer andHoppert [13] model with regard to the lipid monolayerboundary layer, but di¡ers in that it is more advancedwith respect to the protein constituents of the inclusion,which we postulate to be in an organized cytoskeletal-likestructure. Furthermore, the Mayer model (or any othermodel) does not conceive the possibility that organized‘porin-like’ structures for the ingress and egress may bepresent on the inclusion surface.

Acknowledgements

This research was supported by grants from the Nation-al Science Foundation (#0113202 and # 0071717).

References

[1] Anderson, A.J. and Dawes, E.A. (1990) Occurrence, metabolism,metabolic role, and industrial uses of bacterial polyhydroxyalka-noates. Microbiol. Rev. 54, 450^472.

[2] Steinbu«chel, A., Hustede, E., Liebergesell, M., Pieper, U., Timm, A.and Valentin, H. (1993) Molecular basis for biosynthesis and accu-mulation of polyhydroxyalkanoic acids in bacteria. FEMS Microbiol.Rev. 10, 347^350.

[3] Gerngross, T.U., Reilly, P., Stubbe, J., Sinskey, A.J. and Peoples,O.P. (1993) Immunocytochemical analysis of poly-beta-hydroxybuty-rate (PHB) synthase in Alcaligenes eutrophus H16: localization ofthe synthase enzyme at the surface of PHB granules. J. Bacteriol. 175,5289^5293.

[4] Wieczorek, R., Pries, A., Steinbuchel, A. and Mayer, F. (1995) Anal-ysis of a 24-kilodalton protein associated with the polyhydroxyalka-noic acid granules in Alcaligenes eutrophus. J. Bacteriol. 177, 2425^2435.

[5] York, G.M., Stubbe, J. and Sinskey, A.J. (2002) The Ralstonia eu-tropha PhaR protein couples synthesis of the PhaP phasin to thepresence of polyhydroxybutyrate in cells and promotes polyhydroxy-butyrate production. J. Bacteriol. 184, 59^66.

[6] Potter, M., Madkour, M.H., Mayer, F. and Steinbuchel, A. (2002)Regulation of phasin expression and polyhydroxyalkanoate (PHA)granule formation in Ralstonia eutropha H16. Microbiol. SGM148, 2413^2426.

[7] Lundgren, D.G., P¢ster, R.M. and Merrick, J.M. (1964) Structure ofpoly(L-hydroxybuturic acid) granules. J. Gen. Microbiol. 34, 441^446.

[8] Steinbu«chel, A., Aerts, K., Babel, W., Follner, C., Liebergesell, M.,Madkour, M.H., Mayer, F., Pieper-Furst, U., Pries, A. and Valentin,H.E. et al. (1995) Considerations on the structure and biochemistryof bacterial polyhydroxyalkanoic acid inclusions. Can. J. Microbiol.41, 94^105.

[9] Stuart, E.S., Lenz, R.W. and Fuller, R.C. (1995) The ordered macro-molecular surface of polyester inclusion bodies in Pseudomonas oleo-vorans. Can. J. Microbiol. 41, 84^93.

[10] Stuart, E.S., Tehrani, A., Valentin, H.E., Dennis, D., Lenz, R.W. andFuller, R.C. (1998) Protein organization on the PHA inclusion cyto-plasmic boundary. J. Biotechnol. 64, 137^144.

[11] Fuller, R.C., O’Donnell, J.P., Saulnier, J., Redlinger, T.E., Foster, J.and Lenz, R.W. (1992) The supramolecular architecture of the poly-hydroxyalkanoate inclusions in Pseudomonas oleovorans. FEMS Mi-crobiol. Rev. 103, 279^288.

[12] Fuller, R.C. (1999) Microbial inclusions with special reference toPHA inclusions and intracellular boundary envelopes. Int. J. Biol.Macromol. 25, 21^29.

[13] Mayer, F. and Hoppert, M. (1997) Determination of the thickness ofthe boundary layer surrounding bacterial PHA inclusion bodies andimplications for models describing the molecular architecture of thislayer. J. Basic Microbiol. 37, 45^52.

[14] Boatman, E.S. (1964) Observations on the ¢ne structure of sphero-plasts of Rhodospirillum rubrum. J. Cell Biol. 20, 297^306.

[15] Sudesh, K., Gan, Z.H., Maehara, A. and Doi, Y. (2002) Surfacestructure, morphology and stability of polyhydroxyalkanoate inclu-sions characterised by atomic force microscopy. Polym. Degrad.Stab. 77, 77^85.

[16] Schlegel, H.G., Vonbartheld, R. and Gottschalk, G. (1961) Forma-tion and utilization of poly-beta-hydroxybutyric acid by Knallgasbacteria (Hydrogenomonas). Nature 191, 463.

[17] Budashov, I.A., Kurochkin, I.N., Denisov, A.K., Skripniuk, V.V.and Shabanov, T.A. (1997) The use of atomic-force microscopy forthe analysis of microbiological objects. Zh. Mikrobiol. Epidemiol.Immunobiol. 6, 11^15.

[18] Ahimou, F.O., Touhami, A. and Dufrene, Y.F. (2003) Real-timeimaging of the surface topography of living yeast cells by atomicforce microscopy. Yeast 20, 25^30.

FEMSLE 11172 4-9-03

D. Dennis et al. / FEMS Microbiology Letters 226 (2003) 113^119118

[19] Giocondi, M.C., Milhiet, P.E., Lesniewska, E. and Le Grimellec,C. (2003) From cell imaging to molecular manipulation: biolog-ical applications of atomic force microscopy. Med. Sci. 19, 92^99.

[20] Berquand, A., Mazeran, P.E. and Laval, J.M. (2003) In£uence ofvolume and surface properties on phase contrast in tapping modeatomic force microscopy. Surf. Sci. 523, 125^130.

[21] Matsumoto, K., Matsusaki, H., Taguchi, K., Seki, M. and Doi, Y.(2002) Isolation and characterization of polyhydroxyalkanoates in-clusions and their associated proteins in Pseudomonas sp 61^3. Bio-macromolecules 3, 787^792.

[22] Dennis, D.E. (2002) Unpublished results.[23] Murase, T., Iwata, T. and Doi, Y. (2001) Atomic force microscopy

investigation of poly[(R)-3-hydroxybutyrate] lamellar single crystals :Relationship between molecular weight and enzymatic degradationbehavior. Macromol. Biosci. 1, 275^281.

[24] Murase, T., Iwata, T. and Doi, Y. (2001) Direct observation of en-zymatic degradation behavior of poly[(R)-3-hydroxybutyrate] lamel-lar single crystals by atomic force microscopy. Macromolecules 34,5848^5853.

[25] Rossini, C.J., Arceo, J.F., McCarney, E.R., Augustine, B.H., Dennis,D.E., Flythe, M.D. and Baron, S.F. (2001) AFM studies on poly-3-hydroxybutyrate. In: Macromolecular Symposia Advances in Scan-ning Probe Microscopy of Polymers (Tsukruk, V.V. and Spencer,N.D., Eds.). Wiley-VCH, Weinheim, Germany.

[26] McCarnery, E.R. and Augustine, B.H. (1999) Unpublished results.

FEMSLE 11172 4-9-03

D. Dennis et al. / FEMS Microbiology Letters 226 (2003) 113^119 119