Genome shuffling of the nonconventional yeast Pichia anomala for improved sugar alcohol production

Development of efficient colorimetric assay for sugar alcohol screening

Sugar alcohol-producing strains are usually screened and quantified by thin-layer
chromatography (TLC), high-performance liquid chromatography (HPLC) and p-iodonitrotetrazolium violet (INT) methods 24]–26]. However, these methods are time-consuming or suffer from high-cost and are limited
for high throughput screening. Therefore, it is necessary to develop an efficient
screening approach for sugar alcohol-producing microbes.

In our study, a colorimetric method previously applied in trace detection of polyols
27], 28] was developed and optimized for the high throughput assay of sugar alcohols (Additional
file 1: Fig. S1). D-arabitol was selected as a standard for the method construction because it is the
main sugar alcohol product of P. anomala. By optimizing the reaction system, the standardized assay demonstrated a linear
detection range of D-arabitol from 0 to 12 g/L. Although the linear relation was noticeably altered at
20 g/L sugar alcohol, the colorimetric curve was positively related with the sugar
alcohol concentration and could be applied in the preliminary screening (Fig. 1a, b). To analyze effects of the substrate and by-products on sugar alcohol screening,
an interference experiment was performed at different concentrations of glucose and
ethanol (2–30 g/L). The results showed that glucose and ethanol had no interference
in the quantitative analysis of sugar alcohols by the colorimetric method (Fig. 1a), which indicated that the developed assay is highly efficient for the determination
of the content of sugar alcohol in biological samples. To gain a further understanding
of the accuracy, the reference HPLC and the proposed colorimetric methods were applied
to analyze sugar alcohol at different concentration levels. The results showed that
the data measured by the colorimetric procedure agree with those determined by the
reference HPLC method, and a regression line with an R
²
of 0.9673 was obtained (Fig. 1c; Additional file 1: Fig. S1).

Fig. 1. The construction of a colorimetric method for efficient sugar alcohol assay. a The interference test of the colorimetric method under different metabolites. b The standard curve of colorimetric method for D-arabitol detection. c Comparison of the colorimetric method with the HPLC method for sugar alcohol detection
in different fermentation liquors. Data represent the average values of three independent
experiments with deviation varying between 5 and 10% about the mean. d The construction of the P. anomala mutant library by ARTP and UV mutagenesis. The sugar alcohol production was preliminarily
screened by colorimetric assay. The red line represents the sugar alcohol yield of the initial strain P. anomala HP by the colorimetric method.

In this study, a convenient, reliable and low-cost colorimetric assay was developed
for efficient primary screening and selection of strains with high productivity. The
method is highly specific for sugar alcohols and can be performed on crude, non-purified
extracts. The method uses low hazard and inexpensive reagents and requires only commonly
available equipment. Finally, the method is sensitive and highly reproducible. Compared
with HPLC and TLC methods, the colorimetric method facilitated sugar alcohol detection
and made the operation of screening of sugar alcohol-producing strains more convenient.
Although INT is another efficient method for sugar alcohol detection by specific enzyme
catalysis, it is not suitable for high throughput assay because of the complex process
and the expensive substrate p-iodonitrotetrazolium violet 29]. Therefore, the proposed colorimetric assay has clear advantages over the other methods
and can be applied to high throughput screening for different polyol-producing strains.

Development of a rapid hybrid cell selection procedure via FACS analysis

To achieve the efficient screening of hybrid cells without complementary genetic markers,
FACS analysis based on fluorescent dyes was applied. In this process, hybrid cells
are detected by carrying two dyes, and these cells can be analyzed and selected by
FACS.

In this approach, parental protoplasts were prepared and then labeled with fluorescent
dyes Nuclear Green and Nuclear Red, resulting in green and red fluorescence with laser
excitation at 488 and 641 nm, respectively. After fusion, the hybrids were sorted
by FCM, and the results are represented as dot plots (Fig. 2). As the control, strains without staining showed no fluorescence in the R4 gate
(Fig. 2a). The parental strains showed single red and green fluorescence in different gates
based on the staining with fluorescent dyes Red or Green (Fig. 2b, c). Overlap between the fluorescence regions of Green and Red was also observed,
and possible compensation was performed. As shown in Fig. 2d, R3 is the sorting area showing cells that exhibit high intensity fluorescence with
green and red and is identified as potential hybrid cells. In our study, some 2,500,000
protoplasts were rapidly sorted, and 15,300 potential hybrids were selected. Only
approximately 1,000 colonies were found after incubation for regeneration; most protoplasts
were not regenerated, probably because of damage during protoplast preparation, staining
and laser sorting.

Fig. 2. Flow cytometric analysis of the fluorescence distribution after protoplast staining
and fusion. The parent and hybrids with different fluorescent dyes are represented
as dot plots in the figure. Based on the different excitation and emission parameters, the sorting
results were divided into four regions. R2 and R5 detected strains with single Nuclear
Red and Nuclear Green, respectively. R3 detected possible hybrid strains with Nuclear
Red and Nuclear Green. R4 was used as control to detect the blank strains. a Protoplasts of P. anomala without staining; b protoplasts of P. anomala stained by Nuclear Red; c protoplasts of P. anomala stained by Nuclear Green; d double-positive hybrid cells exhibiting high intensity fluorescence for Nuclear Red
and Nuclear Green.

To facilitate screening and identification of the hybrid cells, different genetic
markers were always necessary in previous studies, such as auxotroph 30] and drug resistance 31]. However, a genetic marker, such as auxotroph, affects the physiology and metabolism
of the strain and leads to reduced performance in the production process. Furthermore,
adding genetic markers to the parent strain is a difficult operation for some nonconventional
strains. In this study the fluorescence-activated cell sorting was applied as a useful
method for the hybrid cell selection of P. anomala without the need for genetic markers; in addition, this method is also available
for the genome shuffling of other microbes. It may be possible to apply the technique
for other native strains that are limited by unclear genetic backgrounds or unskilled
genetic operations.

The construction of a mutant library for genome shuffling by random mutagenesis

In the genome shuffling process, the wild type strain is usually treated by the traditional
physical and chemical mutation methods, and the strains with superior performance
are collected to form the parental library for the next step of recursive protoplast
fusion 31], 32]. In this study, a haploid of sugar alcohol-producing P. anomala TIB-x229 5] was first isolated and identified as P. anomala HP. The mutant library was constructed by ultraviolet (UV) and atmospheric and room-temperature
plasma (ARTP) mutagenesis methods to generate genetic diversity. After the mutagenesis
processes, mutants with the maximum sugar alcohol production were selected from approximately
2,000 mutants by colorimetric screening and were then prepared for the next round
of mutation and screening. Through five rounds of continuous mutagenesis, a parent
library with approximately 10,000 mutants was constructed and analyzed by the aforementioned
colorimetric method (Fig. 1d). The sugar alcohol yield of the positive mutants was further confirmed by an HPLC
method, and the four mutants (U-7, U-9, A-4 and A-1) showed clear superiority for
sugar alcohol production. Compared to the initial P. anomala HP, the mutants U-7 and U-9 treated by UV had 7.3 and 8.9% improvement of sugar alcohol
production. The yields of mutants A-4 and A-1 treated by ARTP were increased by 12.3
and 12.9%, respectively (Fig. 3a). These results showed that there was a slight improvement in mutants after several
rounds of traditional mutagenesis. However, the single traditional mutagenesis was
still a time-consuming process for strain engineering because of the low mutation
rate and less diversity.

Fig. 3. Comparison of bioconversion performance between the initial strain, mutants and shuffled
strains. a Comparison of total sugar alcohols production between the initial strain, mutants
and shuffled strains. b–e Comparison of growth condition, glucose consumption, D-arabitol production and ribitol production between initial strain and shuffled strains
GS2-1, GS2-2 and GS2-3. U-: mutants obtained from five rounds of UV mutagenesis of
P. anomala HP. A-: mutants obtained from five rounds of ARTP mutagenesis of P. anomala HP. GS1-: recombinants generated from the first round of genome shuffling. GS2-:
recombinants generated from the second round of genome shuffling. Data represent the
average values of three independent experiments with deviation varying between 5 and
10% about the mean. Asterisk indicates the significant difference in sugar alcohol production at p  0.001 between TIB-x229 and the mutants based on ANOVA statistical test.

Genome shuffling of P. anomala for improved sugar alcohol production

To further improve the performance of sugar alcohol productivity, the mutant strains
(U-7, U-9, A-4 and A-1) with slightly improved performance were collected as the parental
library for the next step of genome shuffling, which is a powerful means for rapid
breeding of improved organisms without knowledge of the detailed genome information.
To achieve the efficient screening for genome shuffling, the developed colorimetric
assay of sugar alcohol and the FACS method were incorporated into the genome shuffling
procedure for our nonconventional yeast P. anomala (Fig. 4).

Fig. 4. The procedure of genome shuffling for improved sugar alcohol production of P. anomala. The process includes six steps, such as mutant library construction, protoplast formation,
fluorescence labeling, PEG induced protoplast fusion, FACS and colorimetric screening.

The protoplasts were processed and fused by a chemical method induced by polyethylene
glycol 33]. After the first protoplast fusion and screening by FACS, approximately 1,000 colonies
with both red and green fluorescence were preliminarily cultivated and assayed for
sugar alcohol production by colorimetric assay. The selected colonies exhibiting improved
performance were further confirmed by HPLC. In the bioconversion process, D-arabitol and ribitol were produced from glucose by P. anomala. Compared to the parental strain P. anomala HP, three recombinants (GS1-1, GS1-2 and GS1-3) exhibited significantly improved
productivity of total sugar alcohols by 19.5, 25.6 and 23.9%, respectively (Fig. 3a). The isolates GS1-2 and GS1-3 were used as the parent population for the following
round of genome shuffling. Similarly, the resulting second-round isolates were further
screened, and three isolates GS2-1, GS2-2 and GS2-3 were selected and evaluated and
showed increased total sugar alcohol production of 46.1, 46.5 and 47.1 g/L, which
was 29.5, 30.6, and 32.3% higher than that of parental strain P. anomala HP, respectively (Fig. 3a). We compared the relative DNA content between the parent strain and the shuffled
strains by DAPI labeling and FCM (Additional file 1: Fig. S2). Compared to the parental strain P. anomala HP and the referenced haploid yeast Saccharomyces cerevisiae BY4741, the wild type TIB-x229, GS2-1, GS2-2 and GS2-3 strains had diploid DNA content.
We assessed the performance and stability of shuffled strains through the bioconversion
of sugar alcohols. For that purpose, bioconversion in sterile water containing 100 g/L
glucose was used to compare the performance of the evolved strains, GS2-1, GS2-2 and
GS2-3, with that of the original strain TIB-x229. Although the overall growth conditions
were the same in all strains, the shuffled strains showed a slightly faster rate of
glucose consumption (Fig. 3b, c). Likewise, the accumulation rate of D-arabitol and ribitol was higher in the shuffled strains. The yield of D-arabitol in the shuffled strains GS2-1, GS2-2 and GS2-3 was 0.29, 0.31 and 0.32 g/g,
which was 11.5, 19.2, and 23.1% higher than that of the original strain P. anomala TIB-x229, respectively (Fig. 3d). The ribitol production in these shuffled strains was 8.46, 11.23 and 10.98 g/L (Fig.
3e), which was also slightly higher than that in the original strain (7.51 g/L). These
results showed that the improvement of the shuffled strains in sugar alcohol production
was due to the accumulation of D-arabitol and ribitol. In this study, two rounds of genome shuffling achieved efficient
gains in sugar alcohol yield. The results further indicated that genome shuffling
is a much more powerful means for breeding improved organisms, especially for those
strains that have undergone classic strain improvement many times.

In recent years, there are also other different reports on sugar alcohols improvement
including metabolic engineering 34], natural screening 5], fermentation optimization 35] and mutation breeding 36]. However, there was not any study reported about improving performance of sugar alcohol-producing
strains by genome shuffling, because there were some obstacles in this process, such
as a lack of efficient sugar alcohol-detection methods and available yeast selective
markers. In our study, we developed the practicable genome shuffling for sugar alcohol-producing
strains by combining the colorimetric assay and fluorescence-activated cell sorting,
which provided a more efficient way for sugar alcohol-producing strain improvement.