Opinion
Plant Cell & Environment
The triose phosphate utilization limitation of photosynthetic rate: out
of global models but important for leaf models
Luke M. Gregory1,2, Alan M.
McClain1,3,4, David M. Kramer1,3,
Jeremy D. Pardo2,4, Kaila E.
Smith1,2,4, Oliver L. Tessmer1,
Berkley J.Walker1,2, Leonardo G.
Ziccardi5, Thomas D. Sharkey1,3
1MSU-DOE Plant research Laboratory, Michigan State
University, East Lansing, MI 48824 US A
2 Department of Plant Biology, Michigan State
University, East Lansing, MI 48824 USA
3 Department of Biochemistry and Molecular Biology,
Michigan State University, East Lansing, MI 48824 USA
4 Plant Biotechnology for Health and Sustainability
Program, Michigan State University, East Lansing, MI 48824 USA
5 Department of Forestry, Michigan State University,
East Lansing, MI 48824 USA
Xiao et al. (2021) present a method for estimating the variability of
estimated parameters of the Farquhar, von Caemmerer, Berry (FvCB) model
of photosynthesis (Farquhar et al., 1980). This model has been very
effective at predicting photosynthetic responses to CO2,
light, and temperature. The original model assumed one of two
conditions: (1) rubisco is saturated with ribulose 1,5-bisphosphate
(RuBP) and so responds to CO2 with Michalis Menten
kinetics (with a competitive inhibitor/ second substrate oxygen)
(rubisco-limited) or (2) rubisco uses RuBP as fast as it is made (RuBP
regeneration-limted). In that case, rubisco activity is determined by
the rate of RuBP regeneration, typically as a result of being
light-limited. But even though photosynthetic CO2assimilation (A ) is light limited, it responds to increasing
CO2 because of suppression of photorespiration.
Carboxylation plus oxygenation stays constant under RuBP regeneration
limited conditions so if oxygenation goes down as CO2increases, carboxylation will go up. The model was expanded to include a
third condition, where RuBP regeneration is limited by how fast
phosphorylated intermediates, primarily triose phosphates, are converted
to end products, thereby releasing phosphate (Sharkey, 1985). This is
usually called “triose phosphate utilization (TPU ) limitation.”
Xiao et al. (2021) limited their analysis to rubisco-limited and
RuBP-regeneration-limited fittings and said that TPU could also
be included. We have tested how inclusion of TPU affects
parameterization of the FvCB model.
The model is most often parameterized by measuring CO2assimilation as a function of CO2 inside the air spaces
of the leaf (Ci ), called anA /Ci curve. Rubisco-limited data points
show a strong response to CO2 while
RuBP-regeneration-limited points show less response but still increase
with increasing CO2. TPU-limited points are
characterized by no response to CO2 and sometimes an
inhibition under increasing CO2. The condition is
further diagnosed by a decline in photosynthetic electron transport
caused by an increase in CO2 or decrease in
O2 (measured by chlorophyll fluorescence analysis). The
TPU limitation is rarely seen at physiological CO2partial pressure and temperature but is very frequently seen when
CO2 is marginally higher than what the plant experienced
during growth, especially if the temperature during the measurement is
marginally lower than the growth temperature. TPU conditions are also
associated with oscillations in photosynthetic rate (Sharkey et al.,
1986), complicating measurements of TPU -limited photosynthesis
rates.
The parameters that can be estimated by the fitting models are the
maximum rate of rubisco turnover (Vcmax ) and the
rate of electron transport (J ) (since the analysis can be done at
limiting light, this need not be Jmax ). Also
estimated are respiration in the light (previously called day
respiration) (RL ) and mesophyll conductance
(gm ). If TPU is considered, this rate of triose
phosphate use (TPU ). We have used equations proposed by Busch et
al. (2018) to include carbon flow out of photorespiration as glycine
(αG ) or serine (αS ).
Some groups have concluded that TPU limitations are likely to be small
and thus constitute an unnecessary complication for modeling
photosynthesis at global scales (Kumarathunge et al., 2019; Rogers et
al., 2021), and Xiao et al. (2021) also left TPU out of their
recent analysis describing Bayesian methods for estimating parameters of
the FvCB model and the uncertainties in those estimates. Given the
obervations of declining A and photosynthetic electron transport
in their data we believe ignoring TPU can lead to errors. We have
systematically explored the consequences of including or ignoringTPU when parameterizing the FvCB model when TPU is
apparent in the data.
We began by re-analyzing the experimental data provided by Xiao et al.
(2021). Four A /Ci curves measured with
rice were provided. In three out of four cases, reverse sensitivity to
CO2 of A was observed and in all four cases,
photochemical yield (measured by chlorophyll fluorescence analysis)
declined at high CO2. These behaviors indicate thatTPU was occurring. The authors specified in their methods section
that they had to wait much longer for stability at the high
CO2 concentrations and the data at high
CO2 was noisy, also an indicator of TPU. We tested the
effect of adding TPU to the analysis.
We converted the most recent version (2.9) of the fitting spreadsheet
that has been provided by Plant Cell and Environment (Sharkey, 2016) to
an R script with a user-friendly interface (Shiny app), see
https://github.com/poales/msuRACiFit.
The script iteratively fits data sets to biochemical models using
rubisco-limited, RuBP-regeneration-limited, or TPU -limited
assumptions, then calculates which process is likely to be rate-limiting
for each data point, thus eliminating the need to assign specific
limiting process to each of the data points.
We then fitted the data supplied by Xiao et al. (2021), first withoutTPU and then with TPU (Figure 1). For all four curves
supplied (only repetitions 2 and 3 are shown in Figure 1), includingTPU in the fitting improved the fit to the data at high
CO2 and this was reflected in a reduction in the sum of
the squared residuals (SSR) (data for repetitions 2 and 3 are given in
Table 1). The reduction in SSRs was much greater than the increase in
degrees of freedom introduced by including TPU as a fitting parameter.
When data points are treated as J -limited but are actually
limited by another process such as TPU , then J is likely
to be underestimated. The estimate of J was higher whenTPU was included in the analysis (Table 1). Our fitting program
could not estimate gm when TPU -limited
points were treated as being J -limited and hit the limit imposed
during fitting of 100 µmol m-2 s-1Pa-1. Because J -limited measurements hold the
most information on mesophyll conductance, the estimate of mesophyll
conductance is affected by fitting without TPU . When TPU is
included it becomes clear how few data points are J -limited and
since J -limited points have the most information aboutgm it becomes clear why gmcan be difficult to measure when A /Cicurves are measured at satuating light. Using high but not saturating
light can increase the amount of J -limited data when estimatinggm (Sharkey, 2019)(see box 1).
Three of the four A /Ci curves had
noticeable discontinuity in the middle of the curves. We reasoned this
was caused by the method used to make the measurements. It is common for
researchers to report A /Ci curves assessed
by measuring at 400 ppm CO2 and then measuring at a
series of declining CO2 concentrations followed by a
jump back to 400 ppm and measuring at a series of increasing
CO2 concentrations. We call this the split method and it
requires that photosynthesis be identical before and after measuring
photosynthesis at ambient CO2, a requirement that often
does not hold in our experience.
We examined the effect of the sequence of CO2concentrations measured during an A /Cicurve and conclude that these measurements should be made by monotonic
increasing (or decreasing) CO2 as opposed to starting at
an ambient CO2 concentration and going down in
CO2, jumping back to the middle and going up in
CO2 (we call this the split method).
We tested split versus monotonic methods with tobacco (Fig. 1 E and F).
The curves did not show an obvious discontinuity but the SSR was higher
for data generated by a split A /Ci curve
than monotonic curve (Table 1) (These SSRs are comparable because the
models used were the same and so the degrees of freedom did not differ.)
Even when the curves do not show an obvious discontinuity in the middle
when measured by the split method, results from “split” experiments
tend to show stronger deviations from continuous fits to models,
suggesting that hysteresis can strongly impact the interpretation.
Moreover, the discontinuity comes at the section of the curve that has
most information on mesophyll conductance and so significantly reduces
confidence in mesophyll conductance values of such split curves.
We conclude that 1. it is important to include TPU when fittingA /Ci curves that show evidence for it; 2.A /Ci curves should be carried out
monotonically. 3. Additional data may be needed depending on how the
fittings are to be used, for example it may be necessary to measure
curves at saturating and also substaurating light to get robust measures
of all parameters. Because of the danger of over fitting, when possible,
parameters should be fixed. For example, if there are independent
measures of mesophyll conductance or light respiration, these can be
supplied and then fixed during fitting. It must be accepted that some
parameters can change within minutes and this biological source of
variance should be considered. Very rapid, monotonicA/Ci curves are likely to be very helpful in
assessing the physiology of photosynthesis just as a high speed shutter
on a camera helps bring things into focus, especially when the subject
is dynamic.
Reporting the parameters of the FvCB model can be helpful for global
modeling, for detecting effects of the environment on photosynthesis,
and changes in specific components of photosynthetic capacity. For large
datasets fitting batches of curves using programs like R can be very
helpful. What is presented expands on part of an earlier R Package
(Duursma, 2015) but now includes TPU. The Shiny app allows users to test
specific hypotheses and can be a convenient way to explore how changing
conditions such as temperature and light affect predicted rates of
photosynthesis.
Please see
https://github.com/poales/msuRACiFitfor how to access and use the R-script and Shiny app used for this work.
Funders: Division of Chemical Sciences, Geosciences and Biosciences,
Office of Basic Energy Sciences of the United States Department of
Energy (Grant DE-FG02-91ER20021).