High saturated fatty acids proportion in Escherichia coli enhances the activity of ice-nucleation protein from Pantoea ananatis


Correspondence: Kun Zhu, Institute of Microbiology Chinese Academy of Sciences, Beijing 100190, China. Tel./fax: +86 10 5747 0235; e-mail: zhuk@im.ac.cn


The ice-nucleation protein (INP) from Pantoea ananatis was expressed in Escherichia coli. INP expression increased the freezing point of the E. coli culture by a few degrees. Deletion of FabH, an important enzyme in fatty acid biosynthesis, significantly inhibited the ice-nucleation activity. Increased unsaturated fatty acids in the fabH mutant cells decreased the ice-nucleation activity. Adding exogenous saturated fatty acids increased both E. coli fatty acid saturation and the ice-nucleation activity. In contrast, adding unsaturated fatty acids exhibited the opposite effects. Furthermore, an E. coli MG1655-fadR strain with high saturated fatty acids content was constructed, in which the INP activity was enhanced by about 17% compared with its activity in the wild-type MG1655 strain.


The term ice nucleation describes the initiation of transition of water from a liquid to solid state. A natural water sample of moderate size normally does not freeze at 0 °C, keeping supercooling state until −2 °C to −15 °C (Kawahara, 2002). While pure water could be cooled to the temperature near to −40 °C before it freezes (Kozloff et al., 1983; Kawahara, 2002). The ice-nucleation protein (INP) confers on cells the ability to nucleate ice crystals at subfreezing temperatures (as high as −2 °C to −5 °C) (Hirano & Upper, 2000). All INPs are comprised of three distinct structural domains: the N-terminal hydrophobic domain, the C-terminal hydrophilic domain, and central repeating domain. The central repeating domain consists of contiguous repeats of 16 or 48 residues with a consensus sequence (Ala-Gly-Tyr-Gly-Ser-Thr-Leu-Thr) and presumably acts as a template for ice crystal formation (Kawahara, 2002). To date, INP has been found in more than 10 bacterial species, typically Gram-negative phyllospheric bacteria, including Pseudomonas, Erwinia, and Xanthomonas (Maki et al., 1974; Tang et al., 2004; Wu et al., 2006; Li et al., 2012). Phytopathogenic bacteria such as Pseudomonas syringae, Erwinia herbicola, Xanthomonas campestris, and Fusarium moniliforme, synthesize INP to initiate nucleation and consequently induce frost damage to their hosts, in order to gain access to nutrients (Morris et al., 2004). INP has received considerable interest for potential applications on cryopreservation, snow making, frozen food preparation, transgenic crop, and cell-surface display system (Li et al., 2009).

Purified INP only exhibited ice-nucleation activity at temperature below −6 °C; nevertheless, whole P. syringae cells exhibited activity at up to −2 °C. The depression of purified INP activity was probably due to involvement of other components such as lipids. INP in P. syringae formed a lipoglycoprotein complex for maximal ice-nucleation activity (Govindarajan & Lindow, 1988; Kozloff et al., 1991). The lipid seemed to play an important role in anchoring or positioning INP to the cell membrane and initiate ice nucleation (Turner et al., 1991). Because fatty acids were essential components of membrane lipid, the composition of fatty acids might affect ice-nucleation activity (Lindow, 1995; Blondeaux et al., 1999).

The synthesis of fatty acids in Escherichia coli is catalyzed by discrete enzymes (Fujita et al., 2007; Zhu et al., 2009): the 3-ketoacyl-ACP synthase III (FabH) is responsible for the initiation of fatty acid elongation, the acyl-CoA synthase (FadD), and acyl-CoA dehydrogenase (FadE) participate in the degradation pathway of fatty acids. The fatty acid degradation repressor (FadR) was first discovered as a repressor of β-oxidation and required for maximal biosynthesis of fatty acids (Nunn et al., 1983).

In this work, membrane fatty acids in E. coli were shown to play an important role on INP activity, and E. coli host cells containing high saturated fatty acids exhibited enhanced INP activity.

Materials and methods

Bacterial strain and culture condition

The E. coli strains used are listed in Table S1, Supporting information. Escherichia coli was grown in Luria–Bertani (LB) medium at 37 °C. If needed, fatty acids (J&K Scientific, Beijing, China) were added to 1 mg mL−1 in the medium. At the same time, Brij-58 (ACROS, Geel, Belgium) was added to 10 mg mL−1. When used, ampicilin (Amp), kanamycin (Km), and spectinomycin (Spec) were added to 100, 50, and 100 μg mL−1, respectively.

Gene cloning and plasmid construction

The INP gene was amplified by PCR using Pantoea ananatis genomic DNA as the template, digested with BamHI/HindIII (Takara, Dalian, China), and inserted into pQE30 vector (Qiagen, Hilden, German). The resultant plasmid pQE30-INP was transformed into E. coli to express the INP protein.

The FabH gene was amplified by PCR using E. coli genomic DNA as the template. The PCR product was digested with KpnI/HindIII (Takara, Dalian, China) and inserted into pBAD43 vector (Guzman et al., 1995). To overexpress FabH, the resultant plasmid pBAD43-fabH was transformed into E. coli and induced with 0.2% l-arabinose.

When necessary, ampicilin and spectinomycin were added to the medium to ensure the stability of pQE30-INP and pBAD43-fabH.

Fatty acid composition analysis

Escherichia coli was grown in 10 mL LB broth and harvested when OD600 nm reached 1.0. If exogenous fatty acids were added, cells were washed with 10 mL water for three times in order to get rid of the exogenous fatty acids. The lipids were extracted according to Bligh & Dyer (1959). Methyl icosanoate (C20 : 0) was added as an internal standard. The fatty acid methyl esters were prepared and quantified using Shimadzu D2000 gas chromatograph, as described by Zhang et al. (2002).

Ice-nucleation activity assay

Ice-nucleation activity of INP was determined by a droplet-freezing method. Escherichia coli cultures were grown to OD600 nm = 1.0. Thirty drops of culture (20 μL each) were placed on a machine with constant temperature. The effective freezing of the samples was evaluated by visual assessment. The minutes at which 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, and 100% of samples froze were recorded. The results were the means of at least three repeats. The statistic t-test was analyzed in graphpad prism 5.

Cold survival evaluation

Freeze-thaw challenges were performed as follows: E. coli cultures were cultivated until OD600 nm reached 1.0 at 37 °C, then incubated at −20 °C for 1 h and thawed at room temperature. The freeze-thaw challenges were repeated for three times. Then, 50 μL diluted samples were spread on LB plates and grown at 37 °C for 24 h. The survival was the ratio of colony number of treated and untreated samples. The results were the means of at least three repeats. The statistic t-test was analyzed in graphpad prism 5.

Construction of MG1655-fadR strain

The fadR mutation was transduced through P1 phage from BW25113 (Baba et al., 2006) to MG1655. The transductants were selected on Km plate and confirmed by PCR.

Results and discussion

The expression of FabH affects INP activity in E. coli

The INP induces a cascade of ice crystal formation at temperatures much warmer than what would occur spontaneously. In this work, the coding sequence of the P. ananatis INP gene was cloned into pQE30 and expressed in E. coli BW25113. The INP activity was presented as ice-nucleating time according to Vali (1971), because INP activity was inversely proportional to ice-nucleating time. Expression of INP enabled the E. coli culture to freeze at T50 (time of 50% samples frozen) of 1.3 min at −9 °C. In contrast, the E. coli culture without INP expression did not freeze until 5.5 min at −9 °C (Fig. 1a), and did not freeze even for 60 min at −4 °C (data not shown). This observation indicated that this P. ananatis INP was expressed and active in E. coli.

Figure 1.

The INP activity in Escherichia coli was determined by droplet-freezing test. All experiments were performed at −4 °C, except (a). When indicated, fatty acids C12 : 0 and C18 : 1 were added to 1 mg mL−1, respectively. (a) BW25113 strains with and without INP. ○: BW25113 w/INP, ●: BW25113, △: BW25113 + C12 : 0, ▲: BW25113 + C18 : 1, □: BW25113-fabH + C12 : 0, ■: BW25113-fabH + C18 : 1; (b) strains with fadD, fadE, and fabH mutations. ○: BW25113 w/INP, ●: BW25113-fadD w/INP, ■: BW25113-fadE w/INP, ▲: BW25113-fabH w/INP; (c) strains with FabH overexpression. ○: BW25113 w/INP, ●: BW25113 w/INP + fabH, □: BW25113-fabH w/INP, ■: BW25113-fabH w/INP + fabH; (d) strains with INP plus exogenous fatty acids. ○: BW25113 w/INP, ●: BW25113 w/INP + C12 : 0, □: BW25113 w/INP + C18 : 1, ■: BW25113-fabH w/INP, △: BW25113-fabH w/INP + C12 : 0, ▲: BW25113-fabH w/INP + C18 : 1; (e) strains with fadR mutation. ○: BW25113 w/INP, ●: BW25113-fadR w/INP, △: MG1655 w/INP, ▲: MG1655-fadR w/INP. Error bars indicated the standard deviation of three independent experiments.

INP is a membrane-bound protein, and purified INP had extremely low ice-nucleation activity (Muryoi et al., 2003). Lindow (1995) showed that INP activity in P. syringae was associated with membrane fluidity. Blondeaux et al. (1999) reported that adding fatty acids, olive oils, and silicone oils could affect membrane fluidity and INP activity. To analyze the effect of fatty acid composition and membrane fluidity on INP activity in E. coli, INP was expressed in the fabH mutant strain, a fatty acid synthesis mutant, and in the fadD and fadE mutant strains, which are fatty acid degradation mutants. As shown in Fig. 1b, the ice-nucleating time was increased to over 4 min (T50) in the fabH mutant, while in the fadD and fadE mutants, the ice-nucleating times were not apparently affected. This implied that deletion of fabH could effectively decrease INP activity.

To further investigate the effect of FabH on INP activity, FabH was overexpressed both in wild-type and in the fabH mutant strain (Fig. 1c). FabH expression shortened the ice-nucleating time (T50 at −4 °C) in the fabH mutant strain from 5.0 to 2.5 min, which was similar to that of the wild type. No apparent change of ice-nucleating time occurred when FabH was introduced into wild-type BW25113. The decrease of ice-nucleating time indicated that overexpression of FabH could restore INP activity.

Cold survival ratio of cells with INP is affected by FabH

Active INPs help cells survive from freeze-thaw stress (Missous et al., 2007). To study the effect of FabH on INP, the cold survival ratios were determined in the fabH mutant strain and wild-type strain. After three freeze-thaw cycles, the survival ratio was 22% for strain BW25113 and 21% for the fabH mutant. The difference was not statistically significant (Table 1). Expression of INP increased the ratios from 22% to 41% in BW25113 and from 21% to 31% in the fabH mutant strain (Table 1). Therefore, INP provided some degree of freeze-thaw stress protection as reported (Missous et al., 2007). The protection effect of INP in the fabH mutant strain was significantly less when compared with the wild-type strain. When fabH gene was introduced to the fabH mutant strain, the survival ratio was restored to 55%, which was not significantly different from 51% in wild type under the same condition (Table 1). The change of survival ratio under freeze-thaw stress demonstrated that the protection was provided by active INP. Therefore, less INP activity due to fabH mutation resulted in less protection.

Table 1. Comparison of survival ratios of strain with or without INP expression
StrainSurvival (%)Statistical significance
wtfabHwt w/INPwt w/INP + fabHfabH w/INPfabH w/INP + fabH
  1. wt, BW25113, fabH, BW25113-fabH. , no comparison; NS, not significant; *, significant (P < 0.05); **, very significant (P < 0.01); ***, extremely significant (P < 0.001).

wt22NS **
fabH21NS *
wt w/INP41 ** NS ***
wt w/INP + fabH51NSNS
fabH w/INP31 * *** **
fabH w/INP + fabH55NS **

fabH mutation increases unsaturated fatty acid content

FabH is an important enzyme in fatty acid biosynthesis pathway (Fujita et al., 2007). The deletion of fabH resulted in a substantial increase (52–69%) in the proportion of unsaturated fatty acid in total fatty acids, particularly the amount of cis-vaccenate (C18 : 1; Fig. 2). Overexpression of fabH effectively restored the ratio of unsaturated fatty acid/saturated fatty acid (UFA/SFA) in the fabH mutant to the level of wild-type strain (Fig. 2, Table 2). In these strains, INP expression did not affect the fatty acid composition; therefore, UFA change was caused by the fabH mutation alone (data not shown). Thus, only data with INP expression were presented in Fig. 2 and Table 2. These data indicated that fabH mutation led to high ratio of UFA/SFA, which might be the direct cause of decreased INP activity.

Table 2. Comparison of UFA/SFA ratios of strain with INP expression
Strain and growth conditionUFA/SFAStatistical significance
wtwt + fabHfabHfabH + fabHwt + C12 : 0wt + C18 : 1fabH + C12 : 0fabH + C18 : 1
  1. All strains carrying pQE30-INP. wt, BW25113, fabH, BW25113-fabH. , no comparison; NS, not significant; *, significant (P < 0.05); **, very significant (P < 0.01); ***, extremely significant (P < 0.001).

wt1.1NS ** NS ***
wt + fabH1.0NS
fabH2.3 ** * * ***
fabH + fabH1.3 *
wt + C12 : 00.9NS *
wt + C18 : 12.1 *** *
fabH + C12 : 01.3 * **
fabH + C18 : 13.1 *** **
Figure 2.

FabH mutation affected Escherichia coli fatty acid composition. image_n/fml12197-gra-0001.png: C16 : 1, image_n/fml12197-gra-0002.png: C16 : 0, image_n/fml12197-gra-0003.png: C18 : 1, image_n/fml12197-gra-0004.png: C18 : 0. Escherichia coli was grown at 37 °C and fatty acids were extracted and detected as described in 'Materials and methods'. Error bars indicated the standard deviation of the mean of three repeats.

Exogenous fatty acid affects UFA/SFA ratio and INP activity

To study the relationship of UFA/SFA ratio and INP activity, exogenous fatty acids were added in the growth medium to adjust the ratio of UFA/SFA in cells. As expected, the UFA/SFA ratio of cells was reduced from 2.3 to 1.3 by adding lauric acid (C12 : 0), and elevated from 2.3 to 3.1 by adding oleic acid (C18 : 1) in the fabH mutant strain (Table 2). Accordingly, the ice-nucleating time at T50 in the fabH mutant was decreased about 18% by adding lauric acid, and increased about 15% by adding oleic acid (Fig. 1d). After adding the oleic acid to the wild-type strain, the UFA/SFA ratio was increased from 1.1 to 2.1 (Table 2), and the INP activity was slightly inhibited by 12% at T50 (Fig. 1d). No apparent effect of lauric acid on UFA/SFA ratio and INP activity in wild type were observed (Table 2 and Fig. 1d). We believed that the addition of exogenous fatty acid in medium caused a change in cell membrane UFA/SFA ratio, resulting in an adjusted INP activity. To exclude the possibility that fatty acid itself might affect culture freezing behavior, the ice-nucleating time of E. coli without INP was determined. The exogenous fatty acid did not show any significant effect under this condition (Fig 1a). Therefore, without INP, fatty acid itself could not have affected the ice-nucleating time of E. coli culture.

To further confirm the effect of exogenous fatty acid on INP, other fatty acids myristic acid (C14 : 0), palmitic acid (C16 : 0), and palmitoleic acid (C16 : 1) were added to change the UFA/SFA ratios of BW25113 and BW25113-fabH. As expected, exogenous SFA lowered cell UFA/SFA ratio and enhanced the INP activity, and exogenous UFA increased cell UFA/SFA ratio and inhibited the INP activity (Fig. S1a and b).

Lindow (1995) reported that INP stability was low in highly fluid membranes. This was in agreement with our conclusion in that membrane with high UFA content was highly fluid compared with membrane with high SFA content. In that early study, membrane fluidity was adjusted by environment temperatures. In our work, membrane fluidity was adjusted by gene manipulation, and by supplementing exogenous fatty acids. Blondeaux et al. (1999) concluded that INP activity in P. syringae was increased by adding exogenous SFA and UFA. In our work, the addition of SFA showed similar stimulating effect on INP activity, but the addition of UFA showed a negative effect. One explanation for the different effect could be that high proportion of SFA enhanced the stability of cell membrane and facilitated the binding of INP to cell membrane. Blondeaux et al. (1999) used olive oil as UFA; however, olive oil is a mixture of UFA and SFA (UFA content is higher than SFA), which makes it difficult to elucidate the effect of UFA on INP activity.

The INP activity is enhanced in fadR mutant

In order to enhance the activity of INP, an E. coli strain with low UFA seems to be a suitable choice. It is known that FadR is not only a repressor of β-oxidation of fatty acid, but also an activator of FabA, which catalyzes the synthesis of UFA in E. coli (Fujita et al., 2007). Indeed, the BW25113-fadR mutant synthesized significantly less UFA than BW25113. The deletion of fadR resulted in a 47% decrease in UFA/SFA ratio (Fig. 3) and a 16% decrease in ice-nucleating time (T50) compared with BW25113 (Fig. 1e). Furthermore, we noticed that the E. coli strain MG1655 had a lower ratio of UFA/SFA ratio (Fig. 3) and higher INP activity than BW25113 (Fig. 1e). Additionally, mutated fadR gene was transferred by P1 phage transduction from the BW25113-fadR strain to MG1655 to obtain the MG1655-fadR mutant strain. The deletion of fadR resulted in a 53% decrease in UFA/SFA ratio in MG1655 (Fig. 3). The ice-nucleating time in MG1655-fadR mutant was shortened by about 17% (T50) compared with that in MG1655 (Fig. 1e). Therefore, MG1655-fadR mutant was the best host strain we tested for INP expression, because it possessed the lowest UFA/SFA ratio.

Figure 3.

The FadR mutation decreased UFA/SFA in Escherichia coli. Error bars indicated the standard deviation of the mean of three repeats. The statistic t-test was performed in graphpad prism 5.0. *Statistical significant (P < 0.05).

Freeze-dried cells of P. syringae have been used commercially in artificial snow making, as a replacement of silver iodide in cloud seeding (Gurian-Sherman & Lindow, 1993). Because P. syringae is a plant pathogen, E. coli might be a better host choice for INP applications.


Our work was supported by the National Natural Science Foundation of China (31170040) and by State Key Laboratory of Microbiology Resource (grant no. 20110603). We thank Arthur Kuan for the help on English writing.