Papers Page
The papers listed below are available for download in Adobe Reader *.pdf format.
You may view and print these files with the freely redistributable Reader from Adobe.
 Taswell C, 1987, Journal of Immunology 138(2):333341.
A solution to the problems
of cytolysis assays with additional applications to other immunological and biochemical
assays.
After cellular immunoassays are compared with classical bioassays, conventional
methods and consequent problems of data analysis for cytolysis assays are reviewed
and a new solution is proposed. This solution incorporates new methods, called doseresponse
surface assays and analysis (DRSA), which estimate cytolytic activity coefficients
on a surface in a threedimensional space with two dose variables (killers and targets)
and one response variable (counts). These new methods based on doseresponse surfaces
are demonstrated to be more informative and reliable than classical methods based
on doseresponse curves. In a test of the methods' robustness (sensitivity of parameter
estimates to changes in the dose levels of the assay design), cytolytic activity
coefficients estimated by DRSA varied by
≤30% over a reduction of three to four orders of magnitude in the dose levels.
This remarkable robustness should be compared with the corresponding figures of
as much as 500% over less than
<1 order of magnitude for previously published results of coefficients estimated
by conventional methods. DRSA is distinguished from replotofplots methods such
as those used for enzyme inhibition assays in biochemistry, and is recommended as
a more efficient method that should replace replotofplot methods now antiquated
by the advent of microcomputers. DRSA can be applied to any experimental system
that requires an activity coefficient to be estimated on a doseresponse surface
in a space of
≥3 dimensions (≥2 dose variables and one response variable), regardless
of the mathematical model and statistical estimators used to analyze the doseresponse
interaction. Finally, DRSA is compared with the methods known as response surface
methodology (RSM), and is described as a new class of methods to be added to those
that constitute RSM.
 Taswell C, 1987, Cell Separation Methods and Selected Applications
4:109145.
Limiting dilution assays for the
separation, characterization, and quantitation of biologically active particles
and their clonal progeny.
This review chapter was published without a separate abstract. Excerpts from the
introduction and conclusion are available on the
LDA Review Page.
 Taswell C, 1984, Journal of Immunological Methods 72(1):2940.
Limiting dilution assays for
the determination of immunocompetent cell frequencies. III. Validity tests for the
singlehit Poisson model.
A statistical method was developed to test the validity of the singlehit Poisson
model in limiting dilution assays used to determine immunocompetent cell frequencies.
Principles of bioassay, validity tests, and the distinction between modeldiscrimination
experiments and parameterestimation assays are reviewed in the Introduction. The
new test derived and then demonstrated with previously published data is intended
to be used for parameterestimation assays based upon the singlehit Poisson model.
It is a family of related χ^{2}, t, and F tests
for deviations from zero of the slopes of weighted least squares regression plots.
These plots regress the logarithms of singledose estimates f_{i}
of the frequency φ
on the total cell doses λ_{i} and f_{i} on the
total cell dose reciprocals 1/λ_{i}, that is, Y_{i}
= ln f_{i} on X_{i} = λ_{i} and Y_{i}
= f_{i} on X_{i} = 1/λ_{i}. The
test discriminates against alternative models with multiplehit/target responsegeneration
processes, a variable number (dosedependent) of false negatives, and a constant
number (doseindependent) of false positives. Its purpose as a test for parameterestimation
assays, though, is to detect deviations from the singlehit Poisson model and not
to select one of these alternative models. Tests for modeldiscrimination experiments
to select or 'prove' an unknown alternative model are considered in light of relevant
literature reviewed in the Discussion.
 Taswell C, 1984, M.Sc. Thesis, Department of Mathematics,
New York University.
Limiting dilution assays for the determination
of immunocompetent cell frequencies. II. Experimental design for singledose assays.
A statistical method was developed to optimize experimental designs for quantal
response limiting dilution assays used to determine the frequencies of immunocompetent
cells. Formulas for the design optimization criteria were derived by incorporating
a prior distribution for the immunocompetent cell frequency into the design optimization
analysis. This prior distribution was used to parameterize the advance information
known about the frequency. The CramerRao minimum variance and uninformative assay
probability were chosen as the optimization criteria. Tables of these parameters,
their corresponding optimum doses, and other experimental design parameters were
computed for various frequency prior distributions. This method was developed in
order to resolve the problems of previous approaches. The values computed for the
optimum efficiency dose and for the minimum error of the frequency estimate are
respectively higher and lower than those obtained by previous authors. Additional
results predict that it is possible to perform singledose assays successfully (with
low uninformative assay probability) at doses approaching the optimum efficiency
dose. (Singledose assays should be used, however, only when the singlehit Poisson
model used to estimate the frequency has already been proven valid by a multipledose
experiment. References are provided for these validity tests.) To optimize the design
of future assays in a sequence, information from past assays in the sequence should
be used with the method and results presented here. The errors of frequency estimates
could then theoretically be reduced to levels lower than those previously considered
attainable.
 Taswell C, 1981, Journal of Immunology 126(4):16141619.
Limiting dilution assays for
the determination of immunocompetent cell frequencies. I. Data analysis.
A statistical method was developed for the analysis of experimental data from limiting
dilution assays. Formulas for the estimation of the frequency of immunocompetent
cells within a test population were derived by the statistical methods of weighted
averaging, likelihood maximization, and χ^{2} minimization.
Equations for the latter 2 were solved by Newton's method of iterative approximation.
Estimates obtained by these methods were found to be more valid than those obtained
by least squares (LS) fitting as judged by the χ^{2} test and
as established by Monte Carlo experiments. χ^{2} minimization
was chosen as the preferable estimation method with maximum accuracy and precision
(minimum bias and variance) for the standard determination of frequencies; likelihood
maximization was used only for the confirmation of results. When data from previously
published experiments were reanalyzed, both results and conclusions were found to
differ significantly from those originally obtained by LS fitting, thus demonstrating
the importance of using proper data analysis methods. In conjunction with the use
of available calculators or microcomputers, the method presented here provides a
simple and rapid procedure for the valid determination of immunocompetent cell frequencies.
 Taswell C, MacDonald HR, Cerottini JC, 1980, Journal of Experimental
Medicine 151(6):13721385.
Clonal analysis of cytolytic
T lymphocyte specificity. I. Phenotypically distinct sets of clones as the cellular
basis of crossreactivity to alloantigens.
The cellular basis of the cytolytic crossreactivity observed in primary allogeneic
(C56BL/6 antiDBA/2 and C57BL/6 antiC3H/He) mixedleukocyte cultures (MLC) was
investigated by analysis of the specificity of clonal progeny derived from individual
cytolytic T lymphocyte (CTL) precursor cells (CTLP) contained within these populations.
A sensitive mixedleukocyte microculture (microMLC) technique was used with limiting
dilution analysis by Poisson statistics to determine the frequency of CTLP reactive
against both specific and thirdparty (P815 and AKRA) target cells, to calculate
the probability that each microMLC was a clone derived from a single CTLP, and
to examine the specificity of each microMLC assayed separately against both target
cells. A total of 287 phenotypically specific, heteroclitic, and crossreactive
microMLC from the 2 different strain combinations were observed with a relative
frequency of 81, 11, and 8%, respectively, and were calculated to have mean clone
probabilities of 90 and 99% when based, respectively, upon the frequencies of CTLP
reactive against the specific and thirdparty target cells. These clones were estimated
to have an approximate size of 6 x 10^{4} cells, which corresponded to roughly
16 cell doublings during the 7 d of culture. 22 clones were successfully subcloned
and in virtually every case, the subclones retained the specificity phenotype of
the original clone from which they were derived. These results provide direct evidence
for three phenotypically distinct sets of CTL as the cellular basis of crossreactivity
in MLC populations assayed against two different target cells.
 Taswell C, MacDonald HR, Cerottini JC, 1979, Thymus 1(12):119131.
Limiting dilution analysis
of alloantigenreactive T lymphocytes. II. Effect of cortisone and cyclophosphamide
on cytolytic T lymphocyte precursor frequencies in the thymus.
A minimal estimate of the frequency of cytolytic T lymphocyte precursors (CTLP)
in the thymus was determined by application of Poisson statistics to limiting dilution
analysis. A mean CTLP frequency of 1/1467 was obtained for C57BL/6 (H2^{b})
thymus cells activated by DBA/2 (H2^{d}) irradiated spleen cells and assayed
against P815 mastocytoma (H2^{d}) target cells. CTLP frequencies were
also obtained for spleen, nylon wool column purified spleen, peripheral blood, and
lymph node cell populations. The effect of in vivo drug treatments on CTLP frequencies
was then examined. Cortisone at 100 mg/kg dramatically increased the CTLP frequency
in thymus by more than 20fold despite a drastic reduction in the number of total
thymus cells. The same cortisone treatment did not affect the CTLP frequency in
spleen. In contrast, cyclophosphamide at 300 mg/kg decreased the CTLP frequency
in spleen by more than 10fold without affecting that in thymus. Cyclophosphamide
at 100 mg/kg did not produce any significant change. A detailed explanation of the
calculation of CTLP frequencies is provided and their validity is discussed.
 Taswell C, McDuffie FC, Mann KG, 1975, Immunochemistry 12(4):339343.
Immunochemical relationships
of the intermediates of prothrombin activation.
The mechanism of prothrombin activation previously reported from this laboratory
(Heldebrant et al., 1973) has been immunochemically verified by analysis
of the precursor, intermediate, and product relationships of the proteins with a
micro Ouchterlony precipitin test. The cross reactions between purified bovine prothrombin,
intermediates of activation, and thrombin were examined with rabbit and chicken
antisera to each of these proteins. In each instance, the derivative was found to
cross react with its parent protein, yet was found to be completely distinct from
any other protein derived from the same cleavage. The results support a mechanism
in which prothrombin (70,000 daltons) is cleaved to yield intermediates 1 (51,000
daltons) and 3 (23,000 daltons); intermediate 1 is then cleaved to yield intermediates
2 (41,000 daltons) and 4 (13,000 daltons); thrombin (39,000 daltons) is finally
produced from intermediate 2.
