FIGURE 4 Phylogenetic comparison of enzyme sequences to select
diverse enzyme homologs for testing. Scale bars denote substitutions per
site. (a), (b), (c), (d), (e), and (f) represent the phylogenetic tree
which was constructed based on HsRbk, EcRbk, HsPrs, PcPrs, HsNampt, and
MrNampt,
respectively.
Zb, Zosterops borbonicus ; Ok, Oncorhynchus kisutch ;Ct ,Clostridium taeniosporum ; Th, Thermoflexus
hugenholtzii JAD2 ; Oa, Ornatilinea apprima ; Ph,Photobacterium halotolerans ; Se, Superficieibacter
electus ; Ra, Ralstonia sp. A12 ; Pb, Pogonomyrmex
barbatus ; Pv, Pogona vitticeps ; Tt, Tepidiphilus
thermophilus ; Tk, Thermococcus kodakarensis ; Mj,Methanocaldococcus jannaschii ; Lh, Limimonas halophila ;
Ct, Clostridium thermobutyricum ; Lv, Lactobacillus
vaccinostercus DSM 20634 ; Ap, Aptenodytes patagonicus ; Cg,Cottoperca gobio ; Mr , Meiothermus rufus ; Su,Sulfurovum sp. FS06-10 ; Cb, Comamonadaceae bacterium ; Rb,Rhodocyclaceae bacterium ; Tb, Thermomonas brevis ; Lp,Lysobacter prati . Visualizations of trees were generated using
the Interactive Tree of Life tool
(https://itol.embl.de/).
After obtaining the NMN biosynthetic pathway enzymes library,
the
most productive enzyme homolog for each step was rapidly identified by
applying the
normalized
screening procedure in a
step-by-step
manner. At first, 10 candidate
homologs
of Nampt were evaluated using the normalized screening. The
complementation GFP fluorescence showed that the expression of
LpNampt,
CgNampt, and ApNampt via CFPS was poor (Figure 5b). In our experience,
the complementation fluorescence (∆F) value below 200 implied that
the
amount of corresponding protein expressed in CFPS was not able to
activate
pathway reaction
successfully,
so these
homologs
were ruled out from calculating
RT/Fvalues. By comparing the RT/F values of the remaining
seven homologs, it was found that the RT/F value of
SuNampt was the highest. Next, the candidate homologs of Prs were
characterized through the normalized screening procedure by assembling
with SuNampt. MjPrs was picked out because it had the highest
RT/F value (Figure 5c). At last, OkRbk was identified as
the most productive Rbk homolog by combining with SuNampt and MjPrs in
the normalized screening (Figure 5d). By using the normalized screening
procedure, the best performing enzyme homolog of each step was
identified from 10 different enzyme variants in 24 h (i.e., a 16-hour
CFPS reaction step for
expressing
enzyme homologs and a 8-hour assay step for the determination of the
expression level and activity of each homolog).
However,
it would take several weeks if 10 enzymes were tested in vivo due
to the constraints of cell growth and laborious genetic manipulations
(Nielsen & Keasling, 2016; Wu et al., 2016). In addition, Santos and
colleagues recently reported the development of
a
method
for screening of
enzyme
mutant libraries, which also utilized the split GFP system
(Santos-Aberturas et al.,
2015).
However, the enzyme mutants were
expressed
using the cell-based approach
in
this
method,
which
meant that the time-consuming cloning, cell culturing, and lysing
processes were still needed, thus limiting the
screening
speed and throughput of this method.
By
contrast, our normalized screening procedure could be expanded easily
into a high-speed and high-throughput format because the enzyme homologs
could be readily obtained from the efficient CFPS reactions.