(658h) Dynamic Transcriptomic Profiling of Scheffersomyces Stipitis Reveals Key Information of Its Gene Regulatory Network at Genome-Scale

Authors: 
Hilliard, M., Auburn University
Wang, J., Auburn University
Jeffries, T., University of Wisconsin-Madison
He, Q. P., Auburn University

Dynamic Transcriptomic Profiling of Scheffersomyces stipitis
Reveals Key Information of Its Gene Regulatory Network at Genome-Scale

Matthew Hilliard1, Thomas Jeffries2,3, Q. Peter He1 and Jin Wang1

1
Department of Chemical Engineering, Auburn University, AL 36849

2Xylome, Madison, WI 53719

3
Department of Bacteriology, University of Wisconsin at Madison, WI 53706

The rapid development of “omics” technologies offers unprecedented
opportunities to help understand cellular metabolism at genome-scale. Among
different “omics” data, RNAseq based transcriptome
profiling has been used to enhance the understanding of the genome-scale
response of the organism to different stimuli1,2.
While these studies have provided insightful findings, they are most often
limited to studying two steady-state conditions. From a control perspective, as
cellular metabolism is a highly complex dynamic system, we hypothesize that the
transient response could offer significantly more information on the system
dynamics, especially the internal control mechanism.

In this work, we use
Scheffersomyces stipitis as
the model system to validate our hypothesis. Scheffersomyces stipitis is an important yeast
species in the field of biorenewables due to its
desired capacity for utilizing xylose3, and it has been recognized
that redox balance plays an important role in xylose fermentation4.
However, there has not been any systems level understanding on how the shift in
redox balance contributes to the overall metabolic shift in S. stipitis to
cope with reduced oxygen uptake. To gain better understanding on how the
cellular metabolism is regulated in response to reduced oxygen supply, we
designed experiments to
obtain its transcriptomic profiles during the dynamic transition from aerobic
growth to oxygen-limited fermentation.

Compared to the analysis of RNAseq
data among different steady-states, analysis of dynamic transcriptomic data
that are time-dependent presents many new challenges. For example, while
independent-sample is a basic assumption in many traditional analysis methods,
the dynamic transient samples are not independent from each other. In this
work, we will discuss how we address some of these challenges through adapting
tools developed for dynamic systems in control engineering and present our findings.
Specifically, compared to the
initial (aerobic) and final (oxygen-limited) steady states, the transient
states revealed significantly larger changes. Detailed examination of
individual genes also confirmed this. Indeed, among all genes
that showed differential expression during the transition, more than half of
them only show differential expression during the transient states, and
returned completely back to their initial steady state when the transition
ended. In addition, when we used an improved genome-scale metabolic network
model (GEM) developed by our group to integrate the transcriptomic data, the fluxes
predicted by the GEM show 45% agreement with steady state transcription levels
in terms of up or down regulations, however, they show 65% agreement when
compared with initial transient expression levels. Finally, by integrating
principal component analysis (PCA) with hierarchical clustering techniques, we
were able to identify clusters of genes that showed similar dynamic responses
to the perturbation, and to determine which metabolic pathways are affected the
most by the change of the oxygen condition.

Our results
confirm that the transient states indeed offer significantly important
information that the steady-states cannot, which can be crucial in understanding
the regulation of cellular metabolism.

(a)

(b)

Fig. 1 Dynamic transcriptomic profiles offer significantly more information than those of steady states. (a) 2D projection of global gene TPM data from all four trials. The circles and stars denote the aerobic and oxygen-limited steady-states, respectively, which are very close to each other; while the triangles denote the transition states, which show significantly larger range of changes. This is confirmed by a specific example of glutamate dehydrogenase (GDH) (b) where steady-states-only values (dashed lines) present a somewhat misleading picture (with mixed up-and down-regulations) compared to the full picture when the complete transition paths are revealed (solid lines).

[1] Ma, J., Zhang, B.-B., Wang, F., Sun, M.-M., Shen,
W.-J., Lv, C., Gao, Z.,
Chen, G.-X. RNA-Seq Analysis of Differentially
Expressed Genes in Rice under Photooxidation. Russ J Plant Physl+,
(2017).  

[2] Rong-Mullins, X., Ayers, M.C., Summers, M.,
Gallagher, J.E.G., Transcriptional Profiling of Saccharomyces cerevisiae Reveals the Impact of Variation of a
Single Transcription Factor on Differential Gene Expression in 4NQO,
Fermentable, and Nonfermentable Carbon Sources. G3-Genes Genom
Genet
, (2018).

[3] Jeffries, T.W., Grigoriev,
I.V., Grimwood, J., Laplaza,
J.M., Aerts, A., Salamov, A.,
Richardson, P.M., Genome sequence of the lignocellulose-bioconverting
and xylose-fermenting yeast Pichia stipitis. Nat Biotechnol, (2007).

[4] Cho, J.-Y., Jeffries, T.W., Pichia Stipitis Genes for Alcohol Dehydrogenase with Fermentative
and Respiratory Functions. Appl Environ Microb, (1998).