Linking Hsp90 function to micro-environmentaL and stocHastic variation in fLoraL organs of IrIs pumIla L

Hsp90 is an environmentally responsive molecular chaperone that was found to play a key role in buffering against genetic and non-genetic perturbations in the model organisms Arabidopsis and Drosophila. Here we analyzed the buffering capacity of Hsp90 against two kinds of non-genetic factors – stochastic noise and micro-environmental variation of floral organ traits in naturally growing Iris pumila plants. We found no statistical association between the endogenous level of Hsp90 and the floral organ radial symmetry produced by stochastic developmental noise. Conversely, floral organ plasticity in response to micro-environmental variation tended to be greater with decrease in Hsp90b isoform expression.


INTRODUCTION
Heat shock protein 90 (Hsp90) is an abundant and ubiquitous molecular chaperone that interacts dynamically with a diverse but specific group of inherently unstable substrates (client proteins), many of which are involved in signal transduction (R i c h t e r and B u c h n e r , 2001; N o l l e n and M o r i m o t o , 2002).Although the molecular interaction network of the Hsp90 chaperone system is still poorly understood (Z h a o and H o u r y, 2007), Hsp90 has been proved to be a very useful folding machine required to switch these proteins on or off (R i c h t e r and B u c h n e r , 2001).By promoting the activity of numerous signaling proteins in many different developmental pathways, Hsp90 plays a key role in regulating multiple cellular and developmental processes, including cell cycle and transcription control, chromatin remodeling, organism development, and responses to environmental stresses (Yo u n g et al.Although essential for the proper folding of its clients under physiological conditions, Hsp90 can also participate in the refolding of other proteins under conditions of denaturing stress (N a t h a n et al., 1997;Yo u n g et al., 2001).Because of its dual role in signal transduction and stress response, Hsp90 is hypothesized to "link developmental program to environmental contingency" (R u t h e r f o r d and L i n d q u i s t , 1998).Specifically, Hsp90 during stress becomes redirected from its usual substrates to other stress-damaged proteins, thereby reducing the signaling activity of its substrates through numerous target pathways.Depending upon the stress intensity and the availability of free chaperones, modulation of the chaperone and target function of Hsp90 can lead to either buffering or expression of the cryptic genetic variation that is accumulated in genomes under normal environmental conditions (R u t h e r f o r d , 2003).
The role of the Hsp90 chaperone system in developmental buffering against genetic and nongenetic perturbations has recently become a topic of increased interest (R u t h e r f o r d and L i n d q u i s t , 1998; Q u e i t s c h et al  2007).Partial inhibition of the Hsp90 function in accessions and recombinant lines of Arabidopsis thaliana produced a number of morphological phenotypes that were found to be specific to underlying genetic backgrounds (Q u e i t s c h et al., 2002).The robust health of most seedlings exhibiting strong phenotypes indicates that the level of impairment of Hsp90 required to uncover specific polymorphisms was lower than that necessary to interfere with its "housekeeping" functions (Q u e i t s c h et al., 2002).Phenotypic analyses of an A. thaliana line with constitutively reduced Hsp90 expression by RNAi targeting revealed that Hsp90 reduction affects an array of quantitative life-history traits, including flowering time and total seed set, and decreases the developmental stability of repeated traits (S a n g s t e r et al., 2007).Recent reports indicating altered environmental sensitivity in Hsp90reduced Arabidopsis lines (revealed by genome-wide expression analyses) have widened the functions of Hsp90 to include the "genesis and maintenance of plastic responses" as well (S a n g s t e r et al., 2007).
The purpose of this study was to investigate the effect of the molecular chaperone Hsp90 on two sources of non-genetic variation in the same set of floral organ traits from a sample of distinct Iris pumila genotypes naturally growing in the wild.Within-flower variation, i.e., deviations from radial symmetry of identical floral organ parts, results from stochastic perturbations (developmental noise) in repeated floral organ parts as they develop within a single floral whorl of the same flower and environment.Environmental variation, i.e., variation in trait expressions among different flowers within the same clonal individual, arises due to heterogeneity of the micro environments experienced by each flower of a given clone.
The main questions addressed were: (1) Is there covariation of floral organ radial asymmetry (RA) in I. pumila?(2) Do the micro-environmental sensitivities of different floral organ traits covariate among themselves?(3) Does the endogenous level of Hsp90 affect plasticity of floral organ traits in response to micro-environmental variation?and (4) Does Hsp90 buffer flower development against stochastic noise?

MATERIAL AND METHODS
The studied species Iris pumila L. is a rhizomatous perennial herb widely distributed in the Deliblato Sands, a sandy area situated about 50 km northeast of Belgrade (44 o 47' 39" N/ 21 o 20' 00" E to 45 o 13' 10" N/ 28 o 26' 08" E).In its natural habitats, the species forms round shaped clones, markedly polymorphic for flower color.Because the flower color distinctiveness of I. pumila clones is specified by segregation at several gene loci, each of the flower color variants commonly found in the population can be regarded as a unique genotype (Tu c i ć et al., 1988).
Like other congeneric taxa, Iris pumila has flowers that display a highly specialized morphology.The perianth (the outer, sterile whorls of a flower) is not differentiated into the calyx (the outermost perianth whorl) and the corolla (the inner perianth whorl).Instead, all of the perianth elements are similar in appearance (color, shape) and hence are collectively termed the tepals.The tepal whorls are radially symmetrical and united towards the base, forming a floral tube.The interior, fertile whorls of the flower consist of the male sex organs or stamens (anthers plus filaments), which are sheltered beneath the overarching style arms (part of the female sex organ) below the stigmatic lip.In I. pumila, the upper surface of the falls is equipped with a beard consisting of short fine hairs.Because all of the tepals from the same whorl are essentially identical in size and shape, such flower morphology is said to be actinomorphic (meaning "ray-formed") or regular.

Flower harvesting and measurement
In April of 2004, at the peak of blooming phase, 10 large clones differing in flower color (distinct genotypes) were randomly chosen over an exposed dune site and marked with plastic pegs.Thereafter, two or three just opened flowers (in the early male stage) were randomly harvested from each of these clones and placed individually in a 0.5-l-bottle of 70% ethanol for storage until dissection.Later, in the laboratory, each flower was cut up at the base of the perianth, and its upper parts, the falls and the standards, were flattened over a transparent glass plate covered with 50% glycerol.Digital images of the floral organs were recorded using an optical scan- ner (HP ScanJet 3800; 600 dpi resolution).Analyses of digital images were conducted using IT software (UTHSCSA Image Tool, Version 3.0, San Antonio, Texas).The floral organ traits analyzed in this study were fall length (FL), fall width (FW), standard length (SL), and standard width (SW), all expressed in mm (Fig. 1).All measurements were performed blind with respect to clone genotype.To reduce measurement error, the size of each floral organ was taken two times from the same digital pictures by the same person (AV).Repeated measurement of the whole sample was conducted after the first measurement was completed.
Most floral traits are quantitative in nature, which means they are influenced by multiple gene loci with small additive effects, as well as by various sources of environmental heterogeneity, including random developmental noise (C o n n e r , 2002).Environmental heterogeneity originates from fluctuations in external environmental conditions, while developmental noise results from the stochastic nature of cellular processes regularly occurring even in a homogeneous environment.In this study, sensitivity to stochastic noise (developmental instability) was measured by radial asymmetry (RA), defined as random deviations away from radial symmetry and expressed in terms of within-flower standard deviation of the repeated floral organ parts.Plasticity (PL) of a floral trait was estimated as the absolute difference between the means of repeated floral parts measured on different flowers from the same clone.This measure is an estimate of the effects of microenvironmental variation on floral trait expressions within a given clone, where environmental robustness decreases as PL increases.

leaf sample collection
Simultaneously with flower harvesting, a fully expanded leaf from two ramets per each clone was collected and immediately frozen in liquid nitrogen.Samples were transported to the laboratory and stored at -70 ºC until preparation.

Tissue extract preparation
Tissue extracts were prepared by pulverization of frozen leaves under liquid nitrogen, followed by sonification on ice in 2 volumes (w/v) of 0.1 M Tris buffer, pH 7.6, containing 1 mM EDTA and 1 mM dithiothreitol.The samples were centrifuged at 12000 x g at 4ºC for 15 min.The supernatants were re-centrifuged and analyzed for total protein content by the method of S p e c t o r (1978) with bovine serum albumin used as a standard.

SDS-pAGE and immunoblotting
Aliquots of tissue extracts containing 40 µg of proteins were mixed with an equal volume of 2X SDSsample buffer, boiled for 5 min at 100ºC, and loaded onto 7.5% SDS-polyacrylamide gels (L a e m m l i , 1970).Proteins were separated by electrophoresis at 120 V and 4ºC using a Mini-Protean II Electrophoresis Cell (Bio-Rad Laboratories, Hercules, CA).Myosin (205 kDa), β-galactosidase (116 kDa), phosphorylase b (97 kDa), bovine serum albumin (66 kDa), and carbonic anhydrase (29 kDa) were used as molecular mass references.
After electrophoresis, Western transfer of proteins from the gels to nitrocellulose membranes (Hybond-C, Amersham, UK) was carried out at 135 mA and 4ºC overnight using a Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad Laboratories, Hercules, CA) in a transfer buffer containing 25 mM Tris buffer, pH 8.3, 192 mM glycine, and 20% (v/v) methanol.The membranes were blocked for 1.5 h at room temperature in PBS (1.5 mM KH 2 PO 4, 6.5 mM Na 2 HPO 4 , 2.7 mM KCl, and 0.14 M NaCl, pH 7.2,) containing 0.25% nonfat dry milk.Isoforms of Hsp90 were detected with the aid of mouse monoclonal antibody SPA-830 (1:2000; StressGen, Canada).Immunoreactive bands were visualized and quantified by an enhanced chemifluorescence (ECF) detection system using STORM Imager and ImageQuant software (Amersham Biosciences Limited, UK).In order to make quantitative comparisons between multiple immunoblots reliable, an internal reference sample, the leaf extract of an I. pumila plant growing independently of the studied populations, was run on each gel.Prior to any comparison, the intensity of each analyzed immunospecific band was normalized to the intensity of the respective internal reference band on the same blot.A representative immunoblot is presented in Fig. 2.

Statistical analyses
Floral measurements from three replicate flowers were taken individually to assess RA and PL or were averaged together within each clone to obtain genotypic means of floral organ traits.Genotypic means for each floral organ trait and the corresponding means for Hsp90 levels were used to estimate the Pearson product moment correlation coefficients.

RESULTS AND DISCUSSION
Flowers of most angiosperms consist of four types of specialized organs (sepals, petals, stamens, and carpels) that are arranged in a series of concentric rings or whorls.Development of these organs is under the control of several classes of regulatory genes (A, B, C, and E), each of which is expressed in spatially restricted positions within the floral meristem (We i g e l and M e y e r o w i t z , 1993; B o w m a n , 1997; K r i z e k and F l e t c h e r , 2005).Apart from class E genes (which specify identity of all organ types), sepal identity is directed by class A genes, petal identity by classes A + B, stamen by classes B + C, while carpel identity is specified by the expression of class C genes.The molecular and genetic mechanisms that control flower development are to a large extent similar between eudicots and monocots, suggesting that all angiosperms employ homologous genes to specify the identity of their floral organ primordia (K r i z e k and F l e t c h e r , 2005).
The flower is a highly evolved plant structure that manifests a considerable degree of robustness to perturbations from genetic and non-genetic sources.This is visible in the relatively constant floral phenotype within populations and the enormous phenotypic diversity among species and higher taxa.
Table 1 summarizes the results of correlation analysis, which mirrors the strength of statistical associations among RAs of four floral organ traits (FL, FW, SL, and SW) in I. pumila.Although all of these traits share the same precursor tissue -the floral meristem -and, as in other monocots, their identities are specified by the combined expression of A + B class genes (see B o w m a n , 1997 for details), none of the correlation coefficients between floral RAs was statistically significant, implying that small random perturbations of cellular processes producing asymmetry of repeated floral parts exerted their effects very locally and separately.
Unlike RAs, plasticities in response to microenvironmental variation of floral traits appeared to be strongly correlated in I. pumila, especially between two length traits: FL and SL (Table 2).Plastic correlations are thought to originate between traits that exhibit similar sensitivity to a given environmental factor, exert the same function, and/or have a shared genetic basis (S c h l i c h t i n g , 1989; J u e n g e r et al., 2005).These comparative plastic responses are  postulated to promote the maintenance of an integrated phenotype in the face of changing environmental conditions (S c h l i c h t i n g , 1989).
Associations between the parts of fully formed morphological structures can originate in two different ways: by direct connections between the developmental pathways that produce these structures, or by parallel variation of separate pathways that respond to the same extrinsic factors (K l i n g e n b e r g , 2005).In the first case, a developmental precursor can be partitioned into two or more descendant organs or tissues, or there can be inductive signaling from one developing part to another.In either of these cases, perturbations that occur during the interaction or variation that has arisen in the common upstream part of the pathway can be shared or passed down through the partitioning step to multiple parts simultaneously (K l i n g e n b e r g , 2005).The main feature of covariation caused by direct interaction of developmental pathways is that variation arising within the pathways themselves is the source of covariation between traits.
In contrast to covariation due to direct connections between the developmental pathways, the mechanism of parallel variation in independent developmental pathways is based on the action of a shared variable step, such as an allelic variation of a gene that is involved in both pathways, or an environmental factor to which both pathways respond simultaneously but separately (K l i n g e n b e r g , 2005).In this case, the factor responsible for the association between traits lies outside the individuals.
Table 3 shows the Pearson correlation coefficients between RA and PL of the same floral traits in I. pumila.Since none of these coefficients approached statistical significance, it suggests that the mechanism maintaining the stability of floral organ phenotypes against stochastic noise is not the same as that buffering environmental variation.
Molecular chaperone proteins belonging to the Hsp90 family are reported to be components of a buffering mechanism against genetic and environmental perturbations in the model organisms Arabidopsis and Drosophila (R u t h e r f o r d and L i n d q u i s t , 1998; Q u e i t s c h et al., 2002; M i l t o n et al., 2003, 2006; S a n g s t e r et al., 2007).In I. pumila, Hsp90 was found to buffer cryptic genetic variation that can be uncovered during environmental stress (M a n i t a š e v i ć et al., 2007).
In spite of significant progress achieved during the last years in elucidating the functions of Hsp90s across biological kingdoms, the relationship between processes buffering against genetic, environmental, and stochastic variation are mostly unidentified, especially in wild plants.To determine whether Hsp90 controls robustness of I. pumila flowers under physiological conditions, we contrasted natural variation in the level of two Hsp90 isoforms with the strength of two kinds of non-genetic perturbations, developmental noise and micro-environmental variation within a wild population of this species.Table 4 presents the magnitude of correlations between the level of expression of Hsp90 isoforms designated Hsp90a and Hsp90b (M a n i t a š e v i ć et al., 2007), and stochastic and micro-environmental variation in four floral organ traits of I. pumila.Our study provides evidence that with the exception of a highly significant positive correlation between RA SW and Hsp90a (r = 0.73; P = 0.02), Hsp90 did not affect the asymmetry of any floral trait, suggesting that it cannot be a general buffer against random perturbations in I. pumila.In contrast to RA, PL of floral organs in I. pumila tended to be negatively related with the endogenous level of Hsp90b, which was previously found to be a strong capacitor of leaf variation in I. pumila (M a n i t a š e v i ć et al., 2007).The impact of Hsp90b on floral plasticity appeared to be stronger in organ lengths than organ widths (r = -0.58 and -0.57 vs. r = -0.43 and -0.37, respectively; Table 4), indicating that Hsp90-dependence of environmental responsiveness in I. pumila is trait-specific.Analyses of Arabidopsis plants with genetically and/or pharmacologically reduced Hsp90 levels established that Hsp90 is essential not only for maintaining the wild-type phenotype against developmental noise and environmental variation, but also for the origin and maintenance of plastic responses to environmental cues (Q u e i t s c h et al., 2002; S a n g s t e r et al., 2007).Previous reports on the participation of Hsp90 in standard plastic responses such as the response to light and gravitropism, together with recent genome-wide expression analyses, clearly indicate that "Hsp90's effects on plant plasticity are generalizable to many different stimuli and that Hsp90 plays a global role in the environmental responsiveness of plants" (S a n g s t e r et al., 2007).
, 2001; Te r a s a w a et al., 2005; R u t h e r f o r d et al., 2007).

Linking Hsp90 function to micro-environmentaL and stocHastic variation in fLoraL organs of IrIs pumIla L.
., 2002; M i l t o n et al., 2003, 2006; D e b a t et al., 2006; M a n i t a š e v i ć Conversely, Hsp90 did protect against the expression of extreme morphogenic phenotypes specific to particular genetic backgrounds or environments, suggesting that genetic and non-genetic buffering by Hsp90 are separable (M i l t o n et al., 2003).It has been recently revealed that Hsp90 effects are remarkably specific to the most invariable and highly discrete quantitative traits, e.g., traits such as the number of scutellar (SC) and thoracic (TH) bristles in Drosophila (M i l t o n et al., 2006).

Table 1 .
Pearson correlation coefficients (P-values in parenthesis) between radial asymmetries (RA) of floral organ traits in Iris pumila; N = sample size.

Table 2 .
Pearson correlation coefficients (P-values in parenthesis) between plasticities (PL) of floral organ traits in Iris pumila; N = sample size.

Table 3
Pearson correlation coefficients (P-values in parenthesis) between radial asymmetry (RA) and plasticity (PL) of floral organ traits in Iris pumila; N = sample size.

Table 4 .
Pearson correlation coefficients (P-values in parenthesis) between Hsp90 (Hsp90a and Hsp90b) expression and stochastic (RA, radial asymmetry) and micro-environmental (PL, plasticity) variation in floral organ traits of Iris pumila.N = sample size.