Two Csp proteins (CspA and CspD) were fused to the green fluorescent protein GFP and expressed from their natural promoters or from an inducible promoter. Fluorescence microscopy and computerized image analysis indicate that in Escherichia coli growing at 37°C CspD localizes in the nucleoid like the control H-NS while CspA occupies a polar position away from the nucleoid. Following cold shock CspA maintains its location, while CspD is not sufficiently expressed to permit its localization. The different localization of CspA and CspD indicates that these proteins play different roles in the cell in spite of their extensive structural similarity.
Following a shift in temperature from 37°C to 10°C (cold shock), Escherichia coli cultures enter an acclimation phase during which gene expression and metabolism undergo changes which permit adaptation of the cells to the new growth temperature. During the growth lag caused by cold shock the synthesis of some proteins increases  and then returns to constitutive levels when the cells resume growth.
The main cold shock protein in E. coli is CspA, a 7 kDa nucleic acid-binding protein [1–3] which stimulates the expression of cold shock genes at the transcriptional [4,5] and translational (; M. Giangrossi and A.M. Giuliodori, manuscript in preparation) levels. In turn, cold shock regulation of cspA occurs transcriptionally, translationally, and also through changes of mRNA stability ( and references therein). CspA could be an RNA chaperone stimulating RNA degradation and binding nascent RNA during transcription, a role actually demonstrated for its homologue CspE. Indeed, in E. coli there are nine highly homologous genes (from cspA to cspI) some of which respond to cold (cspA, cspB, cspG and cspI) or nutritional (cspD) stress. In spite of their structural similarity, there is no evidence as to whether the nine csp gene products in the cell have similar or somehow specialized roles. Although attempts to amplify specific RNA sequences recognized by these proteins yielded disappointing results, an extensive functional redundancy seems unlikely (for review, see ).
A clue in favor of possible functional differences among Csp family members comes from the different cellular localization of CspA and CspD fused with the green fluorescent protein (GFP) presented in this work. As a control, we have confirmed the localization of H-NS, a DNA-binding protein whose association with the nucleoid had been obtained only in cold-shocked cells subjected to cryo-fixation .
Materials and methods
Construction of the hns-gfp fusion
To construct the hns-gfp fusion (Fig. 1A), hns was amplified by PCR using the oligonucleotides 5′Xbahns (5′-CCACTCTAGAATAAGTTTGAGATTACTACA) and 3′Xmahns (5′-AAAGATTCCCGGGTGATCAGGAAATCGTCG) to obtain a fragment (+5 to +455) in which the TAA stop codon was changed to the Glu codon GAA. Following XbaI and XmaI digestion, this fragment was cloned downstream from the lac promoter in pGFP(ASV) (Clontech, USA) giving rise to phns-gfp(ASV) in which the H-NS coding region is fused in frame, through a spacer encoding 11 amino acids, to GFP(ASV), a mutated GFP with a half-life of 110 min. The in vivo expression of this fusion protein was confirmed by Western blot analysis using both anti-H-NS and anti-GFP antibodies (Fig. 1B) while the biological activity of H-NS within the fusion was demonstrated by the capacity of hns-gfp to repress the bgl operon of E. coli TP504 (F−leu B6 serB1203 thi-1 tonA21 lacY1 supE44 zch506::Tn10 zdd230::Tn9 Δhns) .
Construction of the cspD-gfp and cspA-gfp fusions
The cspD-gfp fusion was constructed amplifying a cspD fragment extending from −271 to +319 using oligonucleotides 5′EcoRIcspD (5′-TGATCGAATTCCAGCCAGT) and 3′XmaIcspD (5′-GAAGACCCGGGCGACTGCC GCTT) as PCR primers. The amplified fragment was cloned in pGFP(ASV) digested with XbaI and XmaI and in pTZgfp digested with EcoRI and XmaI to yield pcspD-gfp(ASV) and pTZcspD-gfp, respectively (Fig. 1A). The pTZgfp was constructed by transferring the fragment encoding GFP(ASV) from pgfp(ASV) into pTZ19R digested with SmaI and HindIII. In pTZgfp, the orientation of the region encoding GFP(ASV) is opposite that of the lac promoter and therefore the GFP cannot be expressed. The cspA-gfp fusion (controlled by the lac promoter (Fig. 1A) was constructed amplifying the +120 to +379 fragment of cspA using 5′XbacspA (5′-TCGCCTCTAGACAC ACTTAATTATTAAAGG) and 3′XmacspA (5′-GCAGAGCTCCC CGGGTGGTTACGTTACCA GCCTGCC) as primers. Plasmid pcspA-gfp(ASV) was obtained by cloning this fragment into XbaI and XmaI digested pGFP(ASV). The construct pTZcspA-gfp was obtained cloning into EcoRI and XmaI digested pTZgfp a cspA fragment extending from −170 to +379, which was amplified using 3′XmacspA (see above) and 5′EcoRIcspA (5′-GCGTTGAATTCAAGCCAAC) as primers.
Microscopic observations and image analysis
Two microliters of the desired culture were placed on microscope slides and observed using a Zeiss Axioplan microscope equipped with a 100× UPlanFluor objective, 1.25× optovar and 4′,6-diamidino-2-phenylindole (DAPI) and GFP filters (Zeiss). Fluorescent photographs were obtained with Fujicolor (ASA 1600). The image analysis program used was created by Dr. Hugues Talbot, CSIRO, Australia. Images were segmented in three phases. (1) Preprocessing: registration was performed manually using carefully selected control points in FITC and DAPI images to overlay precisely the images taken at different wavelengths. A top-hat operation  was then performed to eliminate variations in the background. (2) Seed selection: the background was selected automatically by thresholding, using a pixel level determined manually. The location for each bacterium was selected manually and a seed created for each by drawing a line inside the bacterium. (3) Region growing: contours were defined automatically using the algorithm described by Adams and Bischof . This algorithm grows out regions stemming from the seeds, individually and in parallel, one pixel at a time. The process is repeated and stops only when the image is entirely tessellated. Objects defined in this manner were labeled in raster order and rotated. For each object and for each wavelength a column wise sum is made across the width of the object. The result is a vector of data with as many entries as the object is long, measured in pixels. The column-wise sum is taken as the average fluorescence intensity at that wavelength along the length of the bacterium. Data for all bacteria were normalized along the longitudinal axis and this axis was subsequently divided into 10 regions. The amount of fluorescence within each region was expressed as a percentage of the total fluorescence within the bacterium. The mean percent total fluorescence within each region for all bacteria was calculated and these data were represented on histograms.
Results and discussion
In this study we have selected CspA and CspD because they are among the better characterized members of the Csp family whose expression occurs in different moments of cell growth, namely entering (CspD) and exiting (CspA) stationary phase and in response to two different types of stress such as nutritional (CspD) and cold (CspA) . The in vivo localization of these proteins fused to the GFP  was compared to that of H-NS, an abundant DNA-binding protein which had previously been immunologically localized in the nucleoid in cryosubstituted cells ( and references therein). The genes encoding the chimeric fusions of GFP with H-NS, CspA and CspD were prepared as described in Section 2 and placed under the control of the inducible lac promoter (phns-gfp(ASV), pcspA-gfp(ASV) and pcspD-gfp(ASV)) and, in the case of cspA-gfp and cspD-gfp, also under the control of the cspA and cspD natural promoters yielding pTZcspA-gfp and pTZcspD-gfp. The production of the fusion proteins by E. coli JM109 cells harboring the above vectors (Fig. 1A) was ascertained by SDS–PAGE and immunological Western blot analysis (Fig. 1B).
Aliquots of cell cultures taken at different phases of the growth cycle were observed in an epifluorescence microscope without prior fixation to visualize the fusion proteins with respect to the nucleoid which was stained by incubating the cells in the presence of 0.2 mg ml−1 DAPI.
In full agreement with the previous localization of H-NS in the nucleoid , the present experiments carried out with phns-gfp(AVS) (after IPTG induction) demonstrated that H-NS-GFP is located within the nucleoid in cold-shocked cells (not shown) and at all stages (i.e. OD600=0.6, 1.5 and 2.5) of growth at 37°C (Fig. 2A,B). The good overlap between the H-NS-GFP (green) and the DAPI (blue) fluorescence can also be appreciated from the normalized distribution of the fluorescence intensity of the two fluorophores along the longitudinal cell axis (Fig. 3A). In contrast to H-NS-GFP the fluorescence of free GFP expressed from pgfp(ASV) is diffused throughout the control cell (not shown).
The cellular localization of CspD-GFP and CspA-GFP was studied after expressing these fusions from the natural promoters of cspD and cspA, respectively. The CspD-GFP fluorescence clearly overlaps that of the DAPI-stained DNA in cells harboring pTZcspD-gfp (Figs. 2C,D and 3B) suggesting that, like H-NS-GFP, also CspD-GFP is associated with the nucleoid. On the contrary, in the cells harboring pTZcspA-gfp the CspA-GFP fluorescence is predominantly found at the cell poles, in a position clearly different from the central location of the DAPI fluorescence (Figs. 2E,F and 3C). In Fig. 2F the nucleoid is less dense because the cells were taken in early log phase and not in stationary phase where a nucleoid structure is more evident.
However, since the success in the localization of the green fluorescence depended on the expression level of the fusion protein which in turn depended on the activity of the cspA and cspD promoters which varied as a function of the phase of growth or, more generally, on the growth conditions, the localization of the fusion proteins was not possible under all experimental conditions. In particular, while the nucleoid localization of CspD-GFP was clearly seen during late exponential growth, when the cspD promoter is active (Fig. 1B) in agreement with an earlier report , the same fusion protein was not seen in earlier phases of growth (Fig. 1B) or during cold shock (not shown) when this promoter is little or not at all active. On the other hand, experiments performed on aliquots of culture taken during the growth phases of cells harboring pTZcspA-gfp demonstrated that CspA-GFP is expressed at very high level after cold shock and at the beginning of the exponential phase of growth at 37°C while it decreases at later stages (not shown), in full agreement with the data of Brandi et al. .
To overcome the problem of CspD-GFP and CspA-GFP localization under conditions in which their expression level falls below detection, their localization was attempted after triggering their expression from an inducible promoter. Substantial expression of CspA-GFP was obtained in cells induced at mid-exponential growth at 37°C (Fig. 1B, lanes 3 and 4) and also at later stages (not shown) and the cellular localization of CspA was the same as that seen above; similar results were obtained when CspA-GFP was expressed from the inducible promoter at different stages of cold shock (not shown). Unlike with CspA-GFP the CspD-GFP fusion was efficiently expressed from the lac promoter only under the same growth conditions when its physiological expression would occur. Thus, CspD-GFP was expressed little or not at all during cold shock and during early or mid-exponential growth so that its localization was impossible under these conditions. Furthermore, while stationary-state cells resuming growth contain a substantial amount of CspD-GFP, regardless of IPTG induction, the corresponding fluorescence remained diffused in the cell (not shown) and was rapidly diluted by cell doubling; these results indicate that cspD mRNA is subject to some sort of post-transcriptional control which restricts its expression to cells entering stationary phase and that only at this stage the conditions exist for the nucleoid localization of CspD. On the contrary, our experiments confirmed the localization of CspA at the cell poles during all phases of cold adaptation and of growth at 37°C, suggesting that its location does not depend on a particular physiological state of the cell but rather on its particular affinity for a cellular target localized at the poles.
In summary, we have localized in vivo H-NS, CspA and CspD using non-toxic fluorescent markers. Our data indicate that H-NS is associated with the nucleoid, thus extending to viable cells the previous electron microscopic observation . Our study further shows that CspA and CspD, although belonging to the same multigene family, not only are expressed at different times during the growth cycle and in response to different types of stress, but are also localized in different parts of the cell. CspA is localized at the poles of the cell away from the nucleoid and is expressed primarily during the early phase of growth and after cold shock, while CspD is associated with the nucleoid in the late exponential phase of growth.
If bulk transcription indeed occurs at the periphery of the nucleoid, as suggested by an earlier study carried out by high resolution autoradiography , the main localization of CspA in the cellular compartment not occupied by the nucleoid could be consistent with the hypothesis that this protein is a chaperone for nascent RNA. However, more recent data indicate that most RNA polymerase is localized within the nucleoid interior, at least in Bacillus subtilis. The localization of CspA is consistent with the findings that this protein is required during transitions from steady state to resumption of growth and from growth at optimal temperature to cold since it has been proposed that the cell poles are the area where bacteria sense environmental changes  and where oriC is localized immediately after cell division .
Since CspD likely binds single stranded nucleic acids like the other Csp proteins ( and references therein), its nucleoid localization may suggest a role of this protein in binding and perhaps protecting the DNA regions of the chromosome present in the single stranded conformation at the time of entry in stationary phase. Beyond these speculations on their functions, our findings provide a clear-cut indication that CspA and CspD perform different roles in the cell in spite of their superficial similarities and high degree of structural homology. Additional studies are obviously necessary to clarify these roles.
We wish to express our gratitude to Dr. Hugues Talbot (CSIRO, Australia) who has created the image analysis program. This work was partially supported by grants from the Italian C.N.R. Target Project on Biotechnology and MURST (PRIN 1999 ‘Molecular mechanisms of thermal adaptation in bacteria’) to C.L.P. and from CNRS/Université Paris XI (UMR C8621) and ARC 6794 to F.L.H.; R.M.E. was the recipient of a E.C. Biotech fellowship (contract No. BIO4CT975141) and of a Leverhulme Trust Scholarship.