Liskin Swint-Kruse, Ph.D.
Department Chair, SOM-Kansas City, Biochemistry and Molecular Biology
Professor, Biochemistry and Molecular Biology
lswint-kruse@kumc.edu
View full biography >

Major Research Interests

Personalized medicine:  Mutations at nonconserved amino acid positions follow novel substitution rules.

Each patient genome can have as many as 10,000 variations in protein coding regions. To filter for amino acid changes that alter protein function, and thus have potential to be medically-relevant, high-perform­ing computer algorithms are needed. Although many algori­thms have been developed, improve­ments are urgently needed.

The approach of the Swint-Kruse lab is to improve the assumptions that underlie these algorithms. Most algorithms incorporate (i) se­quence alignments of evolutionarily-related proteins (homologs); and (ii) textbook, amino acid substitution “rules” derived from lab experiments. However, these rules were derived from experi­ments that were highly biased towards positions that do not change much during evolution (“conserved”).

In contrast, more than half of the amino acids do change during evolution (“nonconserved”) for most proteins. Further, we have identified a class of nonconserved amino acids that do not follow the textbook rules. Instead, these positions act as functional “rheostats” during evolution: Amino acid changes at these positions provide opportunities to “dial” protein function either up or down by various amounts. In patients, such changes could make an individual more or less susceptible to disease or drug interactions. We have also shown that computer algorithms correctly predict substitution outcomes for conserved positions but perform poorly for rheostat positions. 

More information on rheostat positions can be found in this short video.

We are now assessing the prevalence of rheostatic positions in a range of different protein types and testing hypotheses about how to identify them. In addition, we are exploring the underlying physical properties that give rise to these complex mutational outcomes. Our work incorporates the disciplines of biochemistry, bioinformatics, and biophysics. Together, these studies will be used to improve algorithms for personalized medicine.

This work has been funded by the W. M. Keck Foundation and by the NIH.

Building synthetic bacterial transcription circuitry using engineered repressor proteins.

In biotechnology, bacterial metabolic or synthesis pathways can be controlled using the rules of Boolean logic, via transcription factors that respond to external signals (e.g.the addition of small molecules).  In a collaboration with the Bennett lab at Rice University, we have designed synthetic transcription repressors that create “AND” logic gates.  These were created by mixing-and-matching the DNA binding and regulatory domains of 10 LacI/GalR homologs.  The resulting chimeras all bind the same DNA (lacO) but are allosterically regulated by the small molecules specific to each regulatory domain.  At least 4 chimeras can be co-expressed to create Boolean "AND" logic gates.  For example, if an E. coli expressed LacI along with the TreR and RbsR chimeras, then IPTG AND trehalose AND ribose were required to induce the target gene.  In addition, we demonstrated functionality that can be used to create a "NOT" logic gate.  Experiments are in progress to use transcription activators to build “OR” gates.   

This work has been funded by the NIH.

A Cra-Kinase complex alters regulation of central metabolism of γ-proteobacteria

For γ-proteobacteria, several key processes are regulated by the LacI/GalR homolog “Cra” (Catabolite Repressor Activator protein). We identified a novel interaction between Cra and a kinase that contributes to the bacteria's ability to switch among carbon sources. Results will identify new ways to perturb central metabolism in γ-proteobacteria, which might be exploited to target a select group of enteric bacteria.

Computational Tools and Resources Available for General Use

1. MARS algorithm for inserting new sequences into manually-edited sequence alignments without disturbing the existing alignment.
2. CoEvolution Utilities contains software for creating ensembles of randomly sampled protein sequence alignments. This software also enables (i) ensemble averaging of analysis scores; (ii) parallel implementation of 5 mathematically-divergent, commonly-used co-evolution packages; and (iii) comparisons of scores across all possible significance thresholds.
3. CoEvolution Utilities contains software to determine “Eigenvector centrality” from co-evolution networks. This analysis identifies which positions in a sequence alignment exhibit the greatest total evolutionary constraint from their interactions with all other protein positions, as opposed to the strongest individual constraints that occur among pairs of positions (calculated by co-evolution algorithms).
4. Structure comparisons: RESMAP License Agreement and Register to Download RESMAP . For simple comparisons of protein structure, super-impositions of 3D structures can be useful.  However, if one wishes to compare large regions or more than two structures, 3D overlays quickly become uninterpretable.  A 2D network representation of protein interfaces allows many structures to be compared at an intermediate level of detail.  These analyses can be used to represent inter-domain and inter-subunit interactions, protein-ligand interactions (DNA and small molecule), and long-range, intra-domain amino acid interactions.  Such network representation is useful for (i) monitoring structural changes during molecular dynamics simulations, (ii) comparing alternatively liganded structures, and (iii) comparing homolog structures
5.  The AlloRep database of sequence, mutational, and structural information for LacI/GalR homologs. The LacI/GalR family of transcription regulators is widely used as a model system for developing computer algorithms to model protein evolution or predict mutational outcomes. We have surveyed decades of published literature to collect, categorize, and list citations for: (i) all known LacI/GalR amino acid variants, (ii) most of the available structures, and (iii) a gold-standard sequence alignment for bona fide LacI/GalR repressors. The online database is searchable using SQL.
6. The RheoScale calculator for quantitatively discriminating rheostat, toggle, and neutral protein positions. Complex mutational behaviors can be identified for individual protein positions when a range of amino acid substitutions is considered.  The RheoScale calculator condenses experimental data from multiple amino acid substitutions into a set of scores that describe the overall mutation outcomes for that position.  These analyses can be helpful for analyzing data from deep mutational scanning experiments.  The condensed scores also facilitate correlations of functional data with the structural and bioinformatic analyses that are required to improve mutation predictions needed for personalized medicine.  The RheoScale calculator is encoded into a Microsoft® Excel workbook and is also available as an R script. 
   RheoScale Calculator files in GitHub
7. PYK-SubstitutionOME database of sequence, mutational and structural information for pyruvate kinase homologs. 
   This is a database developed in collaboration with Dr. Aron Fenton and describes biochemical, allosteric, mutational, structural, pathological, bioinformatic and computational information for mutational variants of pyruvate kinase isozymes. This information is contained in the Excel™ workbook entitled "PYKSubstitutionOME". The zip file "PYK_Multiple_Sequence_Alignments" contains (i) a sampled MSA for the PYK whole family and (ii) a sampled MSA for the PYK subfamily that contains the human isozymes.
8. Aldolase variant and sequence database.
   Database of biochemically characterized aldolase substitution variants; rabbit and human sequences and sequence analysis scores. Information is contained in an Excel™ workbook and a fasta-formatted, multiple sequence alignment for the broader aldolase family is in a second file.