A Capillary Mixer with Adjustable Reaction Chamber Volume for Millisecond Time-Resolved Studies by Electrospray Mass Spectrometry

Article abstract of DOI:10.1021/ac0346757

A novel continuous-flow apparatus for on-line kinetic studies of (bio)chemical solution-phase processes by electrospray ionization mass spectrometry (ESI-MS) is described. The device is based on two concentric capillaries. Fluid is released from the inner capillary into the intercapillary space, where it mixes with solution flowing through the outer capillary, thus initiating the reaction of interest. Gas-phase analyte ions are formed near the tip of the outer capillary by pneumatically assisted ESI. This setup allows the mixer to be placed directly within the ion source, thus providing a minimal dead volume of ～8 nL. Time-resolved data can be recorded in both spectral and kinetic modes. In the former case, the position of the inner capillary is fixed at various points, such that entire mass spectra can be recorded for selected reaction times. For experiments in kinetic mode, the mass spectrometer monitors the signal intensity at selected m/z values, while the inner capillary is continuously pulled back, thus providing intensity-time profiles for specific reactive species. A theoretical framework is developed that allows the measured kinetics to be analyzed by taking into account the effects of laminar flow within the reaction capillary. Failure to take these effects into account results in erroneous rate constants. Studies on the demetalation kinetics of chlorophyll reveal that the apparatus can reliably measure rate constants up to at least 100 s^-1. This represents a substantial improvement over previous ESI-MS-based kinetic methods. Spectral mode experiments on the refolding of ubiquitin show the changing proportions of denatured and tightly folded protein subpopulations in solution. When monitored in kinetic mode, the refolding process was found to proceed with a rate constant of 5.2 s^-1.

Full text of DOI:10.1021/ac0346757

Anal. Chem. 2003, 75, 6408-6414
A Ca p illa ry Mix e r w it h Ad ju s t a b le Re a c t io n
Ch a m b e r Vo lu m e fo r Millis e c o n d Tim e -Re s o lve d
S t u d ie s b y Ele c t ro s p ra y Ma s s S p e c t ro m e t ry
De re k J . Wils on a nd La rs Kone rm a nn*
Department of Chemistry, The University of Western Ontario, London, ON, N6A 5B7, Canada
solution-phase reactions.3,4 Following the initiation of a (bio)-
chemical process by mixing of two or more reactants, the kinetics
can be monitored on-line, i.e., by direct injection of the reaction
mixture into the ion source. The relative concentrations of multiple
reactive species can be recorded as a function of time with
extremely high sensitivity and selectivity. Transient intermediates
may be identified based on their mass-to-charge ratio or their MS/
MS characteristics. On-line ESI-MS kinetic studies have been
carried out in a wide range of areas, including bioorganic
chemistry,5,6 enzymology,7-11 protein folding and assembly,12,13 and
isotope exchange experiments in the context of protein confor-
mational dynamics.14-19
The use of ionization techniques other than ESI for on-line
kinetic MS studies has been explored by a number of groups.20-22
Due to its versatility, however, ESI-MS remains by far the most
popular technique for studies of this kind. An alternative approach
for kinetic measurements involves the use of quench-flow
techniques in conjunction with off-line MS analysis.23,24 In quench-
A novel continuous-flow apparatus for on-line kinetic
studies of (bio)chemical solution-phase processes by
electrospray ionization mass spectrometry (ESI-MS) is
described. The device is based on two concentric capil-
laries. Fluid is released from the inner capillary into the
intercapillary space, where it mixes with solution flowing
through the outer capillary, thus initiating the reaction of
interest. Gas-phase analyte ions are formed near the tip
of the outer capillary by pneumatically assisted ESI. This
setup allows the mixer to be placed directly within the ion
source, thus providing a minimal dead volume of ∼8 nL.
Time-resolved data can be recorded in both “spectral” and
“kinetic” modes. In the former case, the position of the
inner capillary is fixed at various points, such that entire
mass spectra can be recorded for selected reaction times.
For experiments in kinetic mode, the mass spectrometer
monitors the signal intensity at selected m / z values, while
the inner capillary is continuously pulled back, thus
providing intensity-time profiles for specific reactive
species. A theoretical framework is developed that allows
the measured kinetics to be analyzed by taking into
account the effects of laminar flow within the reaction
capillary. Failure to take these effects into account results
in erroneous rate constants. Studies on the demetalation
kinetics of chlorophyll reveal that the apparatus can
(3) Lee, E. D.; Mu¨ ck, W.; Henion, J. D.; Covey, T. R. J. Am. Chem. Soc. 1 9 8 9 ,
111, 4600-4604.
(4) Konermann, L.; Douglas, D. J. Methods Enzymol. 2 0 0 2 , 354, 50-64.
(5) Sam, J. W.; Tang, X. J.; Peisach, J. J. Am. Chem. Soc. 1 9 9 4 , 116, 5250-
5256.
(6) Bakhtiar, R.; Hop, C. E. C. A. J. Phys. Org. Chem. 1 9 9 9 , 12, 511-527.
(7) Fligge, T. A.; Kast, J.; Bruns, K.; Przybylski, M. J. Am. Soc. Mass Spectrom.
1 9 9 9 , 10, 112-118.
(8) Zechel, D. L.; Konermann, L.; Withers, S. G.; Douglas, D. J. Biochemistry
1 9 9 8 , 37, 7664-7669.
reliably measure rate constants up to at least 1 0 0 s-1
.
(9) Northrop, D. B.; Simpson, F. B. Bioorg. Med. Chem. 1 9 9 7 , 5, 641-644.
(10) Norris, A. J.; Whitelegge, J. P.; Faull, K. F.; Toyokuni, T. Biochemistry 2 0 0 1 ,
40, 3774-3779.
This represents a substantial improvement over previous
ESI-MS-based kinetic methods. Spectral mode experi-
ments on the refolding of ubiquitin show the changing
proportions of denatured and tightly folded protein sub-
populations in solution. When monitored in kinetic mode,
the refolding process was found to proceed with a rate
(11) Paiva, A. A.; Tilton, R. F.; Crooks, G. P.; Huang, L. Q.; Anderson, K. S.
Biochemistry 1 9 9 7 , 36, 15472-15476.
(12) Lee, V. W. S.; Chen, Y.-L.; Konermann, L. Anal. Chem. 1 9 9 9 , 71, 4154-
4159.
(13) Larson, J. L.; Ko, E.; Miranker, A. D. Protein Sci. 2 0 0 0 , 9, 427-431.
(14) Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1 9 9 3 , 115, 6317-6321.
(15) Simmons, D. A.; Konermann, L. Biochemistry 2 0 0 2 , 41, 1906-1914.
(16) Simmons, D. A.; Dunn, S. D.; Konermann, L. Biochemistry 2 0 0 3 , 42, 5896-
5905.
constant of 5 .2 s-1
.
(17) Lin, H.; Dass, C. Rapid Commun. Mass Spectrom. 2 0 0 1 , 15, 2341-2346.
(18) Maier, C. S.; Schimerlik, M. I.; Deinzer, M. L. Biochemistry 1 9 9 9 , 38, 1136-
1143.
Soon after the advent of electrospray ionization mass spec-
trometry (ESI-MS) in the late 1980s,1,2 it became clear that this
technique has an enormous potential for kinetic studies on
(19) Eyles, S. J.; Dresch, T.; Gierasch, L. M.; Kaltashov, I. A. J. Mass Spectrom.
1 9 9 9 , 34, 1289-1295.
(20) Brivio, M.; Fokkens, R. H.; Verboom, W.; Reinhoudt, D. N.; Tas, N. R.;
Goedbloed, M.; van den Berg, A. Anal. Chem. 2 0 0 2 , 74, 3972-3976.
(21) Ørsnes, H.; Graf, T.; Degn, H. Anal. Chem. 1 9 9 8 , 70, 4751-4754.
(22) Northrop, D. B.; Simpson, F. B. Arch. Biochem. Biophys. 1 9 9 8 , 352, 288-
292.
* To whom correspondence should be addressed. E-mail: konerman@uwo.ca;
http:/ / publish.uwo.ca/ ∼konerman/ .
(1) Bruins, A. P.; Covey, T. R.; Henion, J. D. Anal. Chem. 1 9 8 7 , 59, 2642-
2646.
(2) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science
1 9 8 9 , 246, 64-71.
(23) Miranker, A.; Robinson, C. V.; Radford, S. E.; Aplin, R.; Dobson, C. M. Science
1 9 9 3 , 262, 896-900.
6408 Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
10.1021/ac0346757 CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/11/2003

flow experiments, the reaction is initiated by rapid mixing of the
reactants, followed by mixing with a quenching agent, such as
acid, base, or organic solvent, that abruptly stops the reaction after
a specified period of time. An advantage of that technique is the
possible incorporation of purification steps in situations where
components of the reaction mixture would interfere with the MS
analysis. Quench-flow methods undoubtedly represent a powerful
tool for kinetic studies, but they can be problematic in cases where
reactive species are not stable under the conditions of the
quenched reaction mixture. Also, quench-flow studies are labori-
ous because individual time points have to be measured in
separate experiments.
Of particular importance for studies on a wide range of
chemical and biochemical systems are techniques capable of
providing kinetic data on rapid time scales, i.e., seconds to
milliseconds or even microseconds.25 On-line ESI-MS methods
have been used for characterizing processes with half-lives down
to roughly 30 ms.16 This temporal resolution is orders of
magnitude lower than that obtainable in rapid-mixing experiments
with optical detection.26,27 It therefore appears that there might
still be considerable room for extending the time range that is
accessible to MS-based kinetic techniques.
time by changing the solution flow rate is not advisable because
this may result in artifactual changes of analyte ion abundances.32,33
Reaction capillaries of different length are therefore most com-
monly used for recording spectra at different times points. A
drawback of existing continuous-flow methods is the difficulty of
obtaining intensity-time profiles of selected ions. These kinetic
mode data have to be “pieced together” from multiple measure-
ments carried out with different capillary lengths, in a manner
analogous to quench-flow studies.
Up until now, different experimental methods had to be used
for obtaining millisecond time-resolved MS data in kinetic and in
spectral mode. The present study describes a continuous-flow
mixer with adjustable reaction chamber volume that is capable
of both modes of operation. Data can be recorded in kinetic mode
by continually increasing the distance between the mixer and the
ion source, while monitoring the abundance of selected ions.
Alternatively, spectral mode experiments can be performed by
choosing certain (fixed) reaction chamber volumes, such that
entire mass spectra can be generated for selected time points.
The temporal resolution of this system exceeds that of previous
ESI-MS-based kinetic methods.
On-line kinetic studies can be carried out in two different
modes of operation: (i) In “kinetic mode”, the abundance of one
or more ionic species is monitored as a function of time, e.g., by
monitoring the intensity at selected m/ z values on a quadrupole
mass analyzer. This type of experiment provides detailed intensity-
time profiles for individual reactive species, which allows the
accurate determination of rate constants. Stopped-flow ESI-MS is
a method capable of providing highly accurate data in kinetic
mode.28,29 Unfortunately, this approach requires prior knowledge
of the m/ z value(s) of interest, thus posing a serious limitation
for studies on processes that involve unknown intermediates. (ii)
For experiments carried out in “spectral mode”, entire mass
spectra are recorded for selected reaction times, which allows the
detection and identification of transient intermediates. The use
of stopped-flow ESI-MS for studies in spectral mode is difficult,
because entire mass spectra would have to be recorded on a
millisecond time scale, which poses a challenge even for time-of-
flight instruments or quadrupole ion traps. Experiments in spectral
mode are more easily carried out by using continuous-flow
methods. In contrast to stopped-flow ESI-MS, this approach does
not involve real-time data acquisition; spectral mode data can
therefore be recorded even with slow-scanning mass analyz-
ers.5,12,15,30,31 Usually, the reaction chamber in continuous-flow
studies is a capillary that is mounted between a mixer and the
ESI source. The reaction time is determined by the capillary
dimensions and by the solution flow rate. Controlling the reaction
EXPERIMENTAL SECTION
Chemicals. Chlorophyll a from spinach and bovine ubiquitin
were obtained from Sigma (St. Louis, MO). Distilled grade
methanol and hydrochloric acid were supplied by Caledon
(Georgetown, ON, Canada) and glacial acetic acid was supplied
by BDH (Toronto, ON, Canada). All chemicals were used without
further purification.
Optical Stopped-Flow Measurements. These measurments
were performed on an SFM-4 instrument (Bio-Logic, Claix,
France), using an observation wavelength of 664 nm for monitor-
ing the demetalation of chlorophyll. The two stepper motor-driven
syringes used were advanced at 3.5 mL/ s each, for an instrument
dead time of 3.3 ms. All experiments were carried out at room
temperature (22 ( 1 °C).
On-Line Kinetic ESI-MS Measurements. These measure-
ments were carried out using a custom-built continuous-flow
mixing apparatus that is based on two concentric capillaries
(Figure 1). The reaction of interest is initiated by mixing solutions
from syringes 1 and 2 near the end of the inner capillary. The
plungers of both syringes are advanced simultaneously and
continuously by syringe pumps (Harvard Apparatus, model 22,
Saint Laurent, PQ, Canada). The inner capillary consists of fused
silica (100 ( 1.5 µm i.d., 167 ( 3 µm o.d., Polymicro Technologies,
Phoenix, AZ). Its end is plugged by rapid curing, self-priming
polyimide (HD Microsystems, Parlin, NJ). About 2 mm upstream
from this plug, a ∼80 µm deep notch is cut into the side of the
inner capillary, which allows solution from syringe 1 to be expelled
into the ∼8-µm-wide intercapillary space. The outer capillary
consists of stainless steel (182 ( 2 µm i.d., 356 ( 6 µm o.d., Small
Parts, Miami Lakes, FL) and has a length of 13 cm. The inner
capillary passes through a three-way PEEK union (Upchurch
Scientific, Oak Harbor, WA) and is directly connected to syringe
1. Within the PEEK union, a Flexon sleeve (Alltech, Deerfield,
(24) Houston, C. T.; Taylor, W. P.; Widlanski, T. S.; Reilly, J. P. Anal. Chem.
2 0 0 0 , 72, 3311-3319.
(25) Gruebele, M. Annu. Rev. Phys. Chem. 1 9 9 9 , 50, 485-516.
(26) Knight, J. B.; Vishwanath, A.; Brody, J. P.; Austin, R. H. Phys. Rev. Lett.
1 9 9 8 , 80, 3863-3866.
(27) Roder, H.; Shastry, M. C. R. Curr. Opin. Struct. Biol. 1 9 9 9 , 9, 620-626.
(28) Kolakowski, B. M.; Simmons, D. A.; Konermann, L. Rapid Commun. Mass
Spectrom. 2 0 0 0 , 14, 772-776.
(29) Kolakowski, B. M.; Konermann, L. Anal. Biochem. 2 0 0 1 , 292, 107-114.
(30) Konermann, L.; Collings, B. A.; Douglas, D. J. Biochemistry 1 9 9 7 , 36, 5554-
5559.
(32) Van Berkel, G. J.; Zhou, F.; Aronson, J. T. Int. J. Mass Spectrom. Ion Processes
1 9 9 7 , 162, 55-67.
(31) Sogbein, O. O.; Simmons, D. A.; Konermann, L. J. Am. Soc. Mass Spectrom.
2 0 0 0 , 11, 312-319.
(33) Konermann, L.; Silva, E. A.; Sogbein, O. F. Anal. Chem. 2 0 0 1 , 73, 4836-
4844.
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003 6409

sponding to 0.75 µL/ min was used for the experiments of this
work. Control experiments confirmed that the baseline of the
kinetic experiments is unaffected by the positioning of the inner
capillary. Data in spectral mode are obtained for selected time
points τ by monitoring the entire mass spectrum of the reaction
mixture at fixed values of l.
For demetalation experiments, syringe 1 contained 40 µM
chlorophyll in methanol and syringe 2 contained HCl in methanol
at concentrations ranging from 30 to 100 mM. Both syringes were
advanced at 30 µL/ min for a total flow rate of 60 µL/ min in the
reaction capillary. Ubiquitin refolding studies were carried out by
having syringe 1 filled with 20 µM protein in 46% water, 50%
methanol, and 4% acetic acid. Syringe 2 contained water. The two
syringes were advanced at 20 and 50 µL/ min, respectively, for a
total flow rate of 70 µL/ min. Final solution conditions after mixing
were 14.3% methanol and 1.1% acetic acid.
Safety Considerations. Like any experimental technique that
is based on ESI-MS, the method described here involves the use
of high electric potentials. Precautions should be taken to avoid
contact with the outer capillary of the continuous-flow setup and
with the components of the ion sampling interface of the mass
spectrometer during operation. Syringes 1 and 2, and the syringe
pumps used, should be properly grounded.
Fig u re 1 . Schematic cross-sectional diagram of the experimental
apparatus used in this study for time-resolved ESI-MS experiments.
Syringes 1 and 2 deliver a continuous flow of reactants; mixing of
the two solutions initiates the reaction of interest. The inner capillary
can be automatically pulled back together with syringe 1 (as indicated
by the dashed arrow), thus providing a means to control the average
reaction time τ. Solid arrows indicate the directions of liquid flow. Small
arrows in the electrospray (ESI) source region represent the directions
of air flow. Further explanations are given in the text.
IL) around the inner capillary prevents leaking. The opposite end
of the PEEK union is connected to the outer capillary, while its
third port is connected to an inlet tube that delivers solution from
syringe 2. This solution flows through the intercapillary space until
it passes the mixer. The volume of the mixing region can be
approximated as the ∼8-µm-wide and roughly 2-mm-long inter-
capillary space downstream of the notch, which corresponds to
∼8 nL. For a total liquid flow rate of 60 µL/ min, this results in a
theoretical dead time (mixing time) of ∼8 ms.
THEORY AND DATA ANALYSIS
The analysis of kinetic data obtained in continuous-flow
experiments would be easiest in the hypothetical case of “plug
flow”, characterized by a constant flow velocity throughout the
cross-sectional area of the reaction capillary. In this case, τ would
be identical to the reaction time t. Traditional continuous-flow
studies with optical detection are carried out under turbulent flow
conditions, where constant mixing of fast and slow regions within
the capillary effectively causes all analyte molecules to travel with
a velocity close to vj. Data recorded under these conditions can
be analyzed as if there were plug flow.34-37
For on-line ESI-MS experiments, turbulent flow cannot nor-
mally be attained. This is due to the use of relatively narrow
reaction capillaries, typically having an inner radius R of 100 µm
or less. Commonly used flow rates are in the range of tens to
hundreds of microliters per minute, thus resulting in Reynolds
numbers much smaller than the threshold value of 2000.38 Under
these conditions, the flow within the capillary is laminar, with a
velocity profile v(r) that is given by39
A 3-mm-long Delron block accommodates the end of the outer
capillary. It has an inlet for compressed air and is designed to
provide a collateral gas flow around the reaction mixture that exits
from the capillary outlet. The high-voltage power supply of a triple
quadrupole mass spectrometer (PE Sciex, API 365, Concord, ON,
Canada) is connected directly to the outer capillary. This allows
the production of gas-phase ions at the capillary outlet by
pneumatically assisted ESI. Subsequently, these ions pass through
the differentially pumped interface into the vacuum chamber of
the mass spectrometer. The sprayer voltage was held at 6 kV.
For an average flow velocity vj in the reaction capillary, the
(average) reaction time τ at the ion source is given by
τ ) l/ vj
(1)
v(r) ) vmax(1 - (r2/ R2))
(2)
where l is the length of the reaction capillary, i.e., the distance
between the mixer and the capillary outlet. In contrast to previous
continuous-flow ESI-MS systems,5,16,30 l is variable for the setup
used here; it can be controlled by changing the position of the
inner capillary within the outer capillary. For a typical experiment,
the mixer is initially located within the ESI source (i.e., at the
end of the outer capillary), corresponding to τ ≈ 0. The inner
capillary can be continuously pulled back together with syringe 1
by a stepper motor-controlled mechanism. Experiments can
therefore be carried out in kinetic mode by monitoring the
abundance of selected ions as a function of τ, typically with a dwell
time of 30 ms. This mode of operation is possible because the
Flexon sleeve within the three-way union provides a low enough
friction to allow the continuous withdrawal of the inner capillary,
while ensuring a leak-proof connection. A withdrawal rate corre-
where r represents the radial position within the reaction capillary.
The flow velocity at the center of the capillary, vmax, is twice the
average flow velocity vj. This parabolic velocity profile has a
tendency to distort the measured kinetics by “blurring” the time
(34) Fersht, A. Structure and Mechanism in Protein Science; W. H. Freeman &
Co.: New York, 1999.
(35) Johnson, K. A. Methods Enzymol. 1 9 9 5 , 249, 38-61.
(36) Hartridge, H.; Roughton, F. J. W. Proc. R. Soc. (London) 1 9 2 3 , A104, 376-
394.
(37) Owens, G. D.; Margerum, D. W. Anal. Chem. 1 9 8 0 , 52, 91A-106A.
(38) Konermann, L. J. Phys. Chem. A 1 9 9 9 , 103, 7210-7216.
(39) Probstein, R. F. Physicochemical Hydrodynamics, 2nd ed.; John Wiley &
Sons: New York, 1994.
6410 Analytical Chemistry, Vol. 75, No. 23, December 1, 2003

Fig u re 3 . Simulated kinetic profiles for continuous-flow ESI-MS
experiments. The natural logarithm of the signal intensity C(τ) is
plotted as a function of average reaction time τ for a first-order reaction
with C(t) ) exp(-k t). (A) k ) 10 s-1; (B) k ) 1 s-1. Both panels
show three data sets, representing plug flow (solid circles), laminar
flow without diffusion (solid triangles), and laminar flow with diffusion
(D ) 5 × 10-10 m2/s, open circles).
lines in Figure 2. This is in stark contrast to the distribution
functions predicted by eq 4 that have their maximums at a ) τ/ 2.
Now consider a kinetic process for which the concentration
of a particular reactive species as a function of time t is given by
C(t). Kinetic profiles monitored by the mass spectrometer
represent an average concentration C(τ) that can be calculated
according to
Fig u re 2 . Age distribution functions P(τ, a) plotted vs solution age
a for laminar flow. (A) Capillary length l ) 0.168 cm, corresponding
to an average reaction time of τ ) 0.04 s; (B) capillary length l )
16.8 cm, corresponding to τ ) 4 s. Solid lines are distribution functions
calculated from eq 3, assuming a diffusion coefficient D of zero. The
dotted curves are distribution functions simulated for D ) 5 × 10-10
m2/s. The other parameters used are given in the text. Dotted vertical
lines in both panels indicate a ) τ.
C(τ) ) ∞ C(a)P(τ,a) da
(4)
∫
0
axis, because individual positions l along the reaction capillary
cannot be associated with specific reaction times t. Instead, each
value of l corresponds to a range of reaction times that are spread
around the average value τ. We will now develop a data analysis
strategy that takes into account these distortive laminar flow
effects.
For analyte molecules traveling through the reaction capillary,
the “age” a of each molecule is defined as the time required to
move from the mixing point to the ion source. The probability
that an analyte molecule has an age in the range a ... a + da is
given by P(τ, a) da, where P(τ, a) is the “age distribution function”.
For laminar flow, P(τ, a) can be derived from eq 2; it is given by38
This equation is valid for any age distribution function P(τ, a).
For the laminar flow conditions considered here, substitution of
eq 3 into eq 4 results in
τ2
2
da
a3
∞
C(τ) )
τ/ 2C(a)
(5)
∫
To illustrate the effects of laminar flow on the measured C(τ)
profiles we will consider the simple case of C(t) ) exp(-k t).
Kinetic profiles simulated based on eq 5 are shown as solid
triangles in Figure 3 for k ) 10 s-1 (panel A) and k ) 1 s-1 (panel
B). The logarithmic plots of these profiles have a roughly linear
appearance with an average slope of -0.698k. Also shown are the
corresponding kinetic profiles that would be expected in case of
plug flow (solid circles in Figure 3), which have a slope of -k. A
simple-minded “plug-flow analysis” of the kinetic data, assuming
the reaction time t to be equal to the parameter τ, would therefore
introduce an error of ∼30% in the measured rate constants. For
the current study, an iterative least-squares algorithm was
therefore developed for fitting rate constants to the measured
kinetic profiles based on eq 5. This approach can be employed
for any mathematical C(t) expression.
An important assumption made for the derivation of eq 5 was
that the diffusion of analyte molecules within the reaction capillary
is negligible. We will now explore under what conditions this
approximation is justified. Diffusion continuously changes the
radial position and, hence, the flow velocity v(r) of individual
analyte molecules as they travel along the reaction capillary. This
diffusive mixing has a tendency to counteract the distortive effects
τ2 1
2
P(τ,a) )
for a g τ/ 2
a3
and
P(τ,a) ) 0 for a < τ/ 2
(3)
As expected, this equation predicts an average solution age of
) τ at the ion source. The solid lines in Figure 2 show
a
examples of age distribution functions, calculated from eq 3, for
l ) 0.168 cm and for l ) 16.8 cm (corresponding to τ ) 0.04 s
and τ ) 4 s, respectively). The other parameters used for these
calculated curves reflect the experimental conditions used in this
work, i.e., a liquid flow rate of 65 µL/ min, and a capillary radius
of 91 µm, resulting in an average flow velocity of vj ) 0.042 m/ s.
In the hypothetical case of plug flow, P(τ, a) would be a narrow
peak (δ function) centered at a ) τ, as indicated by the dotted
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003 6411

of laminar flow on the measured kinetics. Previous work has
shown that laminar flow effects are virtually eliminated in the case
of38
τ . R2/ 36D
(6)
The dotted graphs in Figure 2 show simulated age distribution
functions for laminar flow in the presence of diffusion. These P-
(τ, a) curves were calculated by using a numerical method,38
assuming a diffusion coefficient of D ) 5 × 10-10 m2 s-1, which
corresponds to a molecule the size of sucrose (this compound
(MW 342) was chosen as an example to illustrate the behavior of
a small biological molecule). The “noisy” appearance of the
distributions in Figure 2 is due to the use of a random number
generator for simulating the diffusion of individual analyte
molecules. The effects of diffusion are insignificant for small values
of τ, and the age distributions functions obtained under these
conditions are very similar to those expected based on eq 3 (e.g.,
for τ ) 0.04 s, Figure 2A). With increasing τ, more pronounced
deviations between the two curves become apparent (e.g., for τ
) 4 s, Figure 2B). In the limiting case described by relation 6,
P(τ, a) resembles a Gaussian curve, centered at a ) τ (data not
shown).38
The effects of analyte diffusion on the measured kinetics can
be taken into account by using the appropriate simulated age
distribution functions in eq 4. Diffusion is insignificant for rapid
chemical processes that require short experimental time windows.
As an example, Figure 3A shows that, for an exponential decay,
C(t) ) exp(-kt) with k ) 10 s-1, virtually identical kinetic profiles
are obtained for laminar flow in the presence of diffusion (D ) 5
× 10-10 m2 s-1, open circles) and in the absence of diffusion (solid
triangles, calculated from eq 5). For slower processes that require
longer experimental time windows, diffusion is no longer negli-
gible. This is illustrated in Figure 3B for an exponential decay
with k ) 1 s-1. Generalizing the results obtained from these
simulations, we conclude that diffusion does not have to be taken
into account for processes that have essentially gone to completion
within a time window of
Fig u re 4 . Demetalation kinetics of chlorophyll a (m/z 894) in
methanol recorded by ESI-MS at three different concentrations of HCl.
Solid line represent fits based on eq 5, with C(t) ) a exp(-kobst).
displaced by two protons.40-42 This reaction represents a conve-
nient test system, because it allows kinetic measurements by ESI-
MS and by standard optical stopped-flow absorption spectroscopy.
When studied under pseudo-first-order conditions, the rate
constant of the reaction is given by kobs ) k[H+]2. The intrinsic
rate constant k in this expression is known to be strongly solvent-
dependent.41,42
The apparatus depicted in Figure 1 was used for monitoring
the kinetics of chlorophyll demetalation in methanol solution for
acid concentrations ranging from 15 to 50 mM. Figure 4 depicts
three representative kinetic profiles, obtained by monitoring the
intensity of singly charged chlorophyll at m/ z 894. Also shown
are fits to the experimental data based on eq 5, with C(t) ) a
exp(- kobst). The pseudo-first-order rate constants kobs obtained
by ESI-MS were plotted as a function of acid concentration (Figure
5, solid triangles). The open circles in Figure 5 represent kobs
values obtained from control experiments carried out by optical
stopped-flow spectroscopy. There is excellent agreement between
these two data sets throughout the whole range, covering pseudo-
first-order rate constants from about 10 to 100 s-1. The use of
higher acid concentrations to obtain even larger rate constants
was not possible due to the onset of corona discharge in the ion
source region. Nevertheless, it is clear that the temporal resolution
of our novel mixing device exceeds that of other on-line ESI-MS
techniques, which so far allowed rate constants up to ∼25 s-1 to
be measured.16
τ < R2/ 36D
(7)
such that the analysis of kinetic data can be carried out based on
eq 5. For R ) 91 µm and D ) 5 × 10-10 m2 s-1, the value of R2/
36D equals 0.46 s, which roughly corresponds to the conditions
of Figure 3A. Of course, this time window will be more extended
for analytes with smaller diffusion coefficients. Equation 5 will also
be valid for analyzing kinetic processes involving two reactants
with different diffusion coefficients (e.g., the association of a
protein with a small molecule), as long as condition 7 is satisfied
for both species. However, if one of the two analytes is present in
large excess, such that its concentration can be considered
constant, only the diffusion coefficient of the limiting reactant will
have to be taken into account.
The solid line in Figure 5 represents a quadratic fit to the kobs
values measured by ESI-MS, based on the expression kobs
)
k[H+]2. The resulting intrinsic rate constant has a value of k )
0.048 ( 0.002 mM-2 s-1. Within experimental error, this is
identical to the k value of 0.050 ( 0.001 mM-2 s-1 that was
obtained through a quadratic fit to the corresponding optical data
(fit not shown). The kobs values obtained for acid concentrations
of 45 and 50 mM evidently deviate from the expected quadratic
behavior; therefore, these data points were not included for the
fitting procedure. This deviation is likely due to a change in
(40) Heaton, J. W.; Lencki, R. W.; Marangoni, A. G. J. Agric. Food. Chem. 1 9 9 6 ,
44, 399-402.
(41) Berezin, D. B.; Drobysheva, A. N.; Karmanova, L. P. Russ. J. Phys. Chem.
RESULTS AND DISCUSSION
Instrument P erformance in Kinetic Mode. The demetala-
tion of chlorophyll a in acidic solution is a well-characterized
1 9 7 6 , 50, 720-723.
process, during which the central magnesium of the porphyrin is
(42) Mazaki, H.; Watanabe, T. Bull. Chem. Soc. Jpn. 1 9 8 8 , 61, 2969-2970.
6412 Analytical Chemistry, Vol. 75, No. 23, December 1, 2003

Fig u re 5 . Pseudo-first-order rate constants kobs for the demetalation
of chlorophyll a as a function of HCl concentration. Solid triangles
are the results of ESI-MS kinetics that were analyzed based on eq 5,
taking into account laminar flow effects. The solid line is a quadratic
fit to these data (kobs values measured for the two highest acid
concentrations were not considered for the fit). Open circles are data
measured by standard optical stopped-flow spectroscopy. Also shown
are kobs values that were obtained from the ESI-MS data by using a
“plug-flow” data analysis that fails to take into account laminar flow
effects (small solid squares). Error bars are standard deviations,
based on at least four independent measurements for each acid
concentration.
reaction mechanism which has previously been found to take place
at high acid concentrations.41
Fig u re 6 . Refolding of ubiquitin studied in spectral mode. ESI mass
spectra are depicted for average reaction times of (A) τ ≈ 0, (B) τ )
160 ms, and (C) τ ) 2.1 s. The spectrum shown in panel D was
recorded 5 min after initiation of refolding in a manual mixing
experiment with off-line analysis. Notation: 13+ represents protein
ions [ubiquitin + 13H]13+, etc. Panels A-C also show some minor
peaks that presumably correspond to fragmentation products of the
more highly charged protein ions.
Figure 5 also shows the values of kobs that are obtained from
an analysis that neglects the laminar flow profile within the
reaction capillary (“plug-flow analysis”, solid squares). In this case,
the measured kinetics were assumed to have the form C(τ) )
exp(kobsτ), with τ ) l/ vj, as defined above. The rate constants
determined by this method are lower than the actual values by a
factor of 0.69 ( 0.03, which is in excellent agreement with the
value of 0.689 that is expected based on the results of the Theory
and Data Analysis section. These observations confirm that
laminar flow effects have to be taken into account for the accurate
determination of rate constants under the conditions of the current
work, and they attest to the validity of eq 5 as a data analysis
tool.
structure breaks down in acidic solutions containing organic
cosolvents such as methanol, to form an extended “A state” that
has a non-native R-helical structure.50 Folding transitions involving
the A state of ubiquitin have previously been studied by ESI-MS,
albeit not in kinetic experiments.14,51
Here, the refolding of ubiquitin is used as a test reaction to
demonstrate the performance of our continuous-flow setup in
kinetic and in spectral mode. The A state was populated by
exposing the protein to 50% methanol and 4% acetic acid; refolding
was initiated by mixing with an excess volume of water (see
Experimental Section). ESI mass spectra of ubiquitin for different
times after initiation of refolding are depicted in Figure 6. The
initial spectrum, recorded for τ ≈ 0 ms (Figure 6A), shows the
13+ and 12+ charge states as the ions with the highest
abundances. This spectrum is practically indistinguishable from
that of the protein prior to initiation of refolding (data not shown)
and very similar to ubiquitin A state spectra that have been
published previously.14,51 As refolding proceeds, the relative
abundance of highly charged protein ions decreases and that of
P rotein Folding Monitored in Spectral and Kinetic Modes.
ESI-MS has become a standard method for monitoring confor-
mational changes of proteins. In the positive ion mode, gas-phase
proteins generated from tightly folded solution-phase conforma-
tions typically show relatively low charge states. In contrast, ions
formed from unfolded proteins are more highly protonated and
show a wider charge-state distribution. The physical reasons
underlying this relationship between solution-phase protein con-
formation and ESI charge-state distribution are still a matter of
debate43-47 Ubiquitin is a small (8565 Da) protein that is commonly
used as a model system for folding studies.48,49 Its compact native
(43) Chowdhury, S. K.; Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1 9 9 0 , 112,
9012-9013.
(44) Fenn, J. B. J. Am. Soc. Mass Spectrom. 1 9 9 3 , 4, 524-535.
(45) Konermann, L.; Rosell, F. I.; Mauk, A. G.; Douglas, D. J. Biochemistry 1 9 9 7 ,
36, 6448-6454.
(46) Kaltashov, I. A.; Eyles, S. J. Mass Spectrom. Rev. 2 0 0 2 , 21, 37-71.
(47) Grandori, R. J. Mass Spectrom. 2 0 0 3 , 38, 11-15.
(48) Vijay-Kumar, S.; Bugg, C. E.; Cook, W. J. J. Mol. Biol. 1 9 8 7 , 194, 531-
544.
(49) Kakuta, M.; Jayawickrama, D. A.; Wolters, A. M.; Manz, A.; Sweedler, J. V.
Anal. Chem. 2 0 0 3 , 75, 956-960.
(50) Brutscher, B.; Bru¨ schweiler, R.; Ernst, R. R. Biochemistry 1 9 9 7 , 36, 6.
(51) Konermann, L.; Douglas, D. J. J. Am. Soc. Mass Spectrom. 1 9 9 8 , 9, 1248-
1254.
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003 6413

with previous work that has shown ubiquitin to be a two-state
folder, except for those few protein molecules that are affected
by proline isomerization.53 Unfortunately, a direct comparison of
our measured kobs values with refolding rate constants from the
literature is not possible, because previous kinetic studies involved
the use of chemical denaturants that are not easily compatible
with ESI-MS. Also, it is noted that wild-type ubiquitin does not
contain any tryptophan residues or other chromophores that could
serve as optical probes in kinetic refolding experiments. Previous
kinetic studies were therefore carried out on a recombinant protein
variant that contained a non-native tryptophan.53,55,56 Because
chromophores are not required for the ESI-MS-based approach
used here, kinetic studies could be performed directly on the wild-
type protein.
Fig u re 7 . Refolding of ubiquitin studied in kinetic mode for four
selected protein ions. Solid lines are fits based on eq 5.
CONCLUSIONS
We have developed a novel method for millisecond time-
the 8+ and 7+ ions increases (Figure 6B,C). The final spectrum
was recorded off-line 5 min after initiation of refolding in a manual
mixing experiment. It shows the 8+ and 7+ ions as the only
dominant peaks, indicating that virtually all of the proteins have
refolded into a compact conformation (Figure 6D).
resolved studies by ESI-MS. The reaction volume of the described
capillary mixing device is adjustable and can be controlled
automatically. In contrast, previously available continuous-flow
methods involved the use of reaction capillaries with different
(fixed) lengths for controlling the reaction time, which caused
the kinetic experiments to be relatively laborious.30,45 Also, the
reproducible positioning of each of the reaction capillaries within
the ion source represents a potential problem with this earlier
approach. The method described in the current study eliminates
both of these difficulties. Our novel mixing device offers the
unique advantage of allowing ESI-MS-based experiments in two
modes of operation; in “spectral mode”, entire mass spectra can
be recorded for selected time points, and in “kinetic mode”, the
intensity of selected ion signals can be monitored as a function of
the average reaction time τ. The device allows first-order rate
constants up to at least 100 s-1 to be measured reliably, which
represents an improvement over previous time-resolved ESI-MS
methods by at least a factor of 4. A potential concern for
experiments on this time scale is the mixing efficiency of the
reactant solutions in the intercapillary space. However, even the
highest rate constants measured by ESI-MS in this work are in
excellent agreement with the results of control experiments
carried out on a standard commercial stopped-flow instrument,
thus indicating that the mixing efficiency is not a limiting factor.
In summary, it appears that the described method represents a
versatile novel tool for kineticists working in a wide range of
different areas, including enzymology, protein folding, and physical
organic chemistry.
Figure 7 shows data measured in kinetic mode, obtained by
recording the abundance of 13+, 12+, 9+, and 8+ ions as a
function of the average reaction time τ. The diffusion coefficient
of ubiquitin can be estimated to be around D ) 1 × 10-10 m2/ s.52
According to condition 7, this implies that diffusion effects on the
observed kinetics are negligible in an experimental time window
up to τ ≈ 2 s. The data can therefore be fitted based on eq 5
(solid lines in Figure 7). C(τ) expressions for the 13+ and 12+
ions were generated by assuming the model function C(t) ) a1
exp(-kobst) + a0, whereas the analysis of the 8+ and 7+ ions was
based on C(t) ) b1(1 - exp(- kobst)) + b0. The use of an offset in
the exponential decays takes into account the observation that
refolding does not go to completion within the experimental time
window of ∼1.3 s. Instead, the kinetic profiles level off at ∼20% of
their initial values. This effect can be attributed to a subpopulation
of slow-folding ubiquitin molecules that contain non-native cis-
proline isomers.53 Proline isomerization, taking place on the order
of tens of seconds,54 is the rate-limiting step for the refolding of
this subpopulation. Figure 6D indicates that after 5 min these slow-
folding proteins have also attained a compact conformation.
Processes occurring on such a slow time scale are beyond the
range accessible by our rapid mixing setup.
The kobs values obtained from the four fits in Figure 7 are in
close agreement: 5.3 (8+), 5.3 (9+), 5.2 (12+), and 5.1 s-1 (13+),
for an average value of 5.2 s-1. Evidently, the experimental kinetics
are well described by these monoexponential fits. This is in line
ACKNOWLEDGMENT
Financial support for this work was provided by the Natural
Sciences and Engineering Research Council of Canada (NSERC),
the Canada Foundation for Innovation (CFI), and the Provincial
Government of Ontario.
(52) Clark, S. M.; Konermann, L. J. Am. Soc. Mass Spectrom. 2 0 0 3 , 14, 430-
441.
(53) Krantz, B. A.; Sosnick, T. R. Biochemistry 2 0 0 0 , 39, 11696-11701.
(54) Brandts, J. F.; Brennan, M.; Lin, L. Proc. Natl. Acad. Sci. U.S.A. 1 9 7 7 , 74,
4178-4181.
(55) Khorasanizadeh, S.; Peters, I. D.; Butt, T. R.; Roder, H. Biochemistry 1 9 9 3 ,
Received for review June 22, 2003. Accepted September
10, 2003.
32, 7054-7063.
(56) Khorasanizadeh, S.; Peters, I. D.; Roder, H. Nat. Struct. Biol. 1 9 9 6 , 3, 193-
205.
AC0346757
6414 Analytical Chemistry, Vol. 75, No. 23, December 1, 2003