The First Discovery of Desmoglein 4:

The Precision and Accuracy of Dr. Nguyen's Biomedical Research

The skin vitally protects the body of all mammalians. This protection is created by the epidermis, which is composed by layers of skin cells called keratinocytes. These keratinocytes tightly attach to one another by structures call desmosomes. 


At desmosomes, keratinocytes attach to one another via molecules called desmocollin and desmoglein. The completion of The Human Genome Project in April 2003 gave us the ability to read nature's complete genetic blueprint for building a human being and know that, indeed, human have only 6 desmocollins (Dsc1a, 1b, 2a, 2b, 3a, and 3b) and 4 desmogleins (DSG1, 2, 3, and 4).


Dr. Nguyen is the first scientist who discovered desmoglein 4 (DSG4) molecule. He was the first who identified the evidence of desmoglein 4 protein and cloned the molecule's mRNA when he studied a skin disease called pemphigus vulgaris. It is a potential lethal disease in which patients develop antibodies against their own skin cells that make them detach from one another and make big skin blisters (Figure 1). The symptom of this disease can be controlled by drugs such as glucocorticosteroid, but the disease still cannot be treated. Dr. Nguyen wants to know what really cause the disease.

It was believed by some scientists that the disease is caused by antibodies against desmoglein 3 (DSG3) because many patients have antibodies against this molecule and, when scientists used antibodies against desmoglein 3 to inject into neonatal mice, an animal model for the disease, the animal developed skin blisters like patients.


However, when Dr. Nguyen did the same experiment with the genetic mice that did not have desmoglein 3, the mice also developed skin blisters. Pure and simple, he proposed that there was a new target molecule for the disease-causing antibodies and the potential novel molecule could be very similar to desmoglein 3.


In 1999, he use a technique, called Western Blot, of proteins isolated from genetic mice lacking of desmoglein 3 and use affinity purified antibodies against desmoglein 3 protein as the probe, he found the first evident of desmoglein 4 protein. The molecule indeed has the size about 130-140 Kd (similar to desmoglein 3). (Nguyen et al, Journal of Clinical Investigation. 2000 (12):1467-79.)


The next step Dr. Nguyen did was to identify the actual identification of this novel molecule. He was one of the first scientists who have used the results of The Human Genome Project to discover new molecules.


The Human Genome Project - one of the great feats of exploration in history-is an international research effort to sequence and map all the genome of human. The project completed in April 2003, however, much earlier than that, since early 2002, a large amount of data from this project has been submitted to the GenBank and therefore available to the public. These data include most of the DNA sequence of human chromosome 18.


With these data in hand, to discover a novel molecule and its transcript mRNA sequence, unlike in the old time, today scientists can do a short cut by starting with a computer and some gene predicting softwares to predict the possible mRNA sequence of a new genes.


Indeed, since early 2002, using some computer softwares such as Genescan and Twinscan, the NCBI Annotation Project (National Center for Biotechnology Information, USA) predicted several potential novel genes such as cadherin 22, 23, and 24 etc, and submitted them to Genbank that available to the public. The sequences that were predicted by these softwares at that time, however, could be inaccurate or predict the genes that actually did not exist.


With the predicted sequence, scientists could prove if the molecule actually exists by simple laboratory experiments such as reverse transcribed polymerase chain reaction (RT-PCR) that is very efficient with the specific and unique sequence of each potential novel molecule.


Since the computer predicted sequence can be wrong, a claim of a novel molecule is only valid if the author(s) can provide actual laboratory data such as the photography of the experimental results and the real critical sample of sequencing electropherogram that shows that they actually did the experiment; and most importantly, the data must be reproducible by independent scientists resulting an identical mRNA sequence of the novel molecule.


In early 2002, Dr. Nguyen began to search for a novel molecule that is similar to desmoglein 3. “Almost every evening, I had dinner in front of my computer, entered the NCBI website, and used a tool called Basic Local Allignmeent Search Tool (BLAST) to search for any new DNA sequence that appeared to be similar to desmoglein 3…” he said.


He found a computational predicted sequence that was very similar to Desmoglein 3. Using the BLAST tool to the genome sequence, he knew that it was predicted from the DNA sequence between the locations of desmoglein 3 and desmoglein 1 genes in chromosome 18q12.1. He called this potential novel molecule desmoglein 4 (DSG4).


He used RT-PCR to clone DSG4 but at first he was able to clone only two third of the molecule at the 5’ end. It appeared that that the computational predicted Desmoglein 4 sequence was not correct at the 3’ end region therefore the primers designed from the data based on computational-predicted DSG4 sequence did not give any results.


Then he down loaded the whole chromosomal DNA sequence of this region as well as some other small EST clones that had sequences similar to Dsg3 and used a technique that he called “Silico cloning” (or cloning by computer) to find other possible mRNA sequences of desmoglein 4. With new sequences, he was able to clone the rest of desmoglein 4 molecule by RT-PCR (Figure 2). 




By sequencing the RT-PCR products (Figure 3), Dr. Nguyen identified the actual sequence of human Desmoglein 4. The resulted sequences constructed a 3531bp cDNA containing a complete open reading frame of 3180 nucleotides encoding for a deducted 1059 amino acid polypeptide with predicted molecular weight of 115.5 kDa for non-glycosylated molecule. Removing of the leading peptide from the site for proteolytic cleavage (RRQKR) predicts a mature non-glycosylated molecule with MW of 109.5kDa (which is similar to MW of non-glycosylated Desmoglein 3, 107.5kDa, that explains why the two molecules co-migrate at the same position in electrophoresis gel).




He submitted the sequence to the GenBank in October 2002 and his record was assigned theaccession number AY168788.1. It is the first laboratory cloned desmoglein 4 sequence has ever been published. It appeared that the actual sequence of human Desmoglein 4 is significantly different from the computational-predicted sequence in the region at 3’ end, which is the portion of the molecule that carries the conserved signaling sequence. It differs of 117 bases that encode for 39 amino acids. RT-PCR using primers that were designed based on the different region of the computational-predicted DSG4 mRNA sequence did not give any result.


With the correct mRNA sequence of the novel desmoglein 4 (DSG4), Dr. Nguyen used the BLAST tool to define the exact 15 exons and 14 introns of desmoglein 4 gene and their locations in the chromosome 18q12.1. The result is different from the computational-predicted DSG4 sequence that showed 16 exons and 15 introns (Figure 4).




The order of arrangement of desmoglein cluster genes, DSG2 – DSG3 – DSG4 – DSG1 in chromosome 18 is probably not a coincidence but rather a result of evolution. It most likely represents the order of protein expression according to the differentiating levels of skin cells (keratinocytes); i.e; DSG2 expresses at the lowest layer of the epidermis, DSG3 and DSG4 express at higher keratinocyte layers, while DSG1 expresses at highest/most differentiated keratinocyte layers (Figure 5).




It is not surprised that there are some differences in the laboratory data of the DSG4 protein expression in skin found in different publications. The DSG4 publication on journal Cell shows that DSG4 express all over the suprabasal keratinocyte layers of the epidermis; while those in some other publications showed only the expression of DSG4 on the top-most differentiated layer. The reason is because different techniques were used in these publications:


RNA In-situ hybridization (hybridization histochemistry) technique uses labeled oligonucleotides specific for DSG4 mRNA sequence therefore it stains any cells where DSG4 mRNA expresses. When the technique was done in parallel with indirect immunofluorescence using DSG4 polyclonal antibody, the result is very reliable (results from DSG4 publication on journal Cell 2003; 113(2): 249-60).


The DSG4 polyclonal antibody is very specific and sensitive to the DSG4 protein in natural form because they were generated by immunizing animal with a large portion of DSG4 that would fold more correctly in the immunized animals. In addition, the antibodies recognize several epitopes at different presentation therefore very sensitive (results from DSG4 publication on journal Cell 2003; 113(2): 249-60).


On the other hand, DSG4 monoclonal antibody could be very specific but less sensitive to the DSG4 protein in natural form. They were usually generated by immunizing animals with a small peptide (only few amino acids - epitope) of DSG4 that was linked to synthetic carriers. Therefore the presentation of the complex in the immunized animals could not be the same as natural DSG4 protein. In addition, even in the case a large portion of the molecule was used for immunization, the monoclonal antibody is generated by a single clone of antibody making cells, it can recognize only one epitope presented in one specific presentation, thus much less sensitive to the targeted protein. 


Since the expression of DSG4 protein in the epidermis is dynamic according to the level of cell differentiation, only at certain presentation where the molecule can expose the epitope that is presented the same specific way as if it was presented by the carrier in immunized animal, it could be recognized by DSG4 monoclonal antibodies (results from DSG4 publication in other journals).


Dr. Nguyen also analyzed the sequence of desmoglein 4 and found several interesting features. Overall, DSG4 shares 79%, 68% and 48% similarity to DSG3, DSG2, and DSG1 respectively. But at the N-terminal regions of these molecules, DSG4 shares a much higher level of similarity to DSG3 and DSG1: 89% and 92% respectively. Along these similar extracellular domains of the molecules, there are several identical amino acid stretches that could be the binding sites (epitopes) for disease causing antibodies in pemphigus (Figure 6). This explains the cross activity of disease causing antibodies against DSG3 and DSG4 from the results of Dr. Nguyen’s experiments in 1999 when he first found the evident of desmoglein 4.


Note that desmoglein 4 and desmoglein 3 have similar sizes. Mature non-glycosylated DSG4 and DSG3 molecules have similar molecular weight. Both proteins have the same glycosylated sites at their extracellular domain 1 and extracellular anchor domain.




Unlike Desmoglein 3, Desmoglein 4 contains a cytoplasmic putative conserved signaling domain typical of classical cadherins (Figure 7). This consensus signaling region could function to induce clustering (sending signal into the cell to cluster proteins necessary to do cell function at the site such as desmosomes). It could bind to the signaling proteins protein such as p120ctn well as others that have been known to perform intracellular signal transduction. 




Since the disease causing antibodies in pemphigus, at the cellular level, cause the disease by inducing signals to the cells that disrupt normal biological activities of skin cells. These signals cause internalization of desmoglein 3 (and probably other cadherins such as desmocollin 3), retraction of cytoskeleton filaments causing unsupported desmosomes, etc…that ultimately lead to cells detach from one another, which is called acantholysis. 


The signaling feature of desmoglein 4 domain at the 3’ end may explain a part of the mechanism for the pathogenesis of the disease pemphigus. Hypothetically, the normal signals from the molecule control the clustering and organization of adhesion-related molecules and therefore control the dynamic of desmosomes during keratinocyte differentiation. Binding of disease causing antibodies to the molecule (as well as some other molecules) induce abnormal signals therefore disrupt this normal process and hence, cause the disease.


Preliminary screening experiments showed that majority of pemphigus patients have antibodies against desmoglein 4. Results from preliminary in vivo experiments showed that antibodies from pemphigus patients against desmoglein 4 can cause pemphigus-like lesions in neonatal rat model (first developed by Dr. Nguyen). This is a better model for pemphigus than the classic neonatal mice because genetically with regard to desmogleins, similar to human, rat has 4 types of desmogleins, 1-4; while mice has 2 additional desmoglein variants. Because of this important difference, the presence of 6 types of desmogleins in mice desmosomes, results of past neonatal mice model for pemphigus might be inaccurate. Some scientists might say the finding about the pemphigus pathogenicity of antibodies against desmoglein 4 is just a cross reactivity with desmoglein 3 and 1, or vice versa. But the important point is this discovery lets us understand better how the skin works and how the skin gets trouble.

The complicate pathogenesis of pemphigus at cellular and molecular levels gives scientists the opportunity to use it as a model to explore the nature of skin and see that skin is tough and that only one hit cannot knock it down. Using antibodies from pemphigus patients as the tool to discover novel human molecules, Desmoglein 4 is the third human molecules (both proteins and genes- others are an annexin and a nicotinic acetylcholine receptor) that Dr. Nguyen discovered before the end of human genome project. 


In 2000 Dr. Nguyen proposed the "multiple hits" theory for the pathogenesis of pemphigus to explain the disease at the cellular and molecular levels. The theory postulates that acantholysis in PV results from simultaneous and cumulative effects of autoantibodies directed toward different keratinocyte self-antigens, including the “structural” antigens, such as desmosomal cadherins, and “functional” antigens, such as cell surface receptors regulating function of the adhesion and cytoskeletal units. Desmoglein 4 acts as both "structural" and "functional" antigen that adds more flavor to the theory. (Nguyen et al, Journal of Biological Chemistry 2000; 275(38): 29466-76.)


Dr. Nguyen’s book chapter, CHAPTER 19 “Experimental Mouse Model of Pemphigus Vulgaris: Passive Transfer of Nondesmoglein 1 and 3 Antibodies“ in the book “Animal Models of Human Inflammatory Skin Diseases” (Editors: Dr. Lawrence S. Chan, Published by CRC in 2003) is the first and single publication that reports the first record of laboratory cloned desmoglein 4, the GenBank AY168788. In this chapter, section D.1 “Identifying the novel pemphigus antigen DSG4“ describes how DSG4 was discovered. As mentioned before, the most important requirement to determine the validity of a scientific claim is its data must be reproducible by independent scientists resulting an identical solid results.


Also in this book chapter, he proposed an hypothetical model how desmocollins and other desmogleins interact with desmoglein 4 to send the signals through Wnt/wingless pathway to control keratinocyte proliferation and differentiation, a process called "direct desmoglein-activated signaling"


In January 31, 2007, a group of 83 scientists lead by Dr. Strausberg at the National Cancer Institute, Bethessda, MD, USA submitted to the GenBank their desmoglein 4 mRNA sequence (GenBank: BC132907.1). Their sequence confirms 100% accuracy of the sequence that Dr. Nguyen reported in October 2002.


The desmoglein 4 mRNA sequence is now has an NCBI Reference Sequence: NM_001134453.1. It is called Homo sapiens desmoglein 4 (DSG4), transcript variant 1, mRNA. In this reference, base 107-3637 that encodes complete DSG4 protein is 100% identical to bases 1-3531 of the sequence from GenBank: AY168788.1. (Figure 8). And of course, the protein sequence is 100% identical to the DSG4 protein sequence reported by Dr. Nguyen in October 2002 (Figure 9).






In the record of the NCBI Reference Sequence: NM_001134453.1, the record refers to 7 publications on the topic of desmoglein 4. However, all these 7 publications reported a different sequence: Homo sapiens desmoglein 4 (DSG4), transcript variant 2 (NCBI Reference Sequence: NM_177986.2). The mRNA sequence in this reference derived from GenBank: AY177664.1 that was first submitted to GenBank in November 2002 by Drs. Whittock and Bower in UK.


These scientists claimed in their publication in the Journal of Investigative Dermatology (issue 120 (4), 523-530, 2003) that they have used RT-PCR to clone the desmoglein 4 mRNA. Using their clone, Dr. Whittock also collaborated with another laboratory in Japan for additional research project that lead to another publication: Nagasaka, et al, J. Clin. Invest. 114 (10), 1484-1492 (2004).


It appears that that the sequence of Desmoglein 4 reported in AY177664.1 is identical to the computational-predicted sequence of the molecule. Also in their publication, they showed no original laboratory evident of the result leading to their sequence. 




Desmoglein 4 (DSG4), transcript variant 1 (NCBI Reference Sequence: NM_001134453.1)

Original data:

1)- GenBank: AY168788.1. Submitted to GenBank in October 2002

2)- GenBank: BC132907.1 Submitted to GenBank in January 2007


Desmoglein 4 (DSG4), transcript variant 2 (NCBI Reference Sequence: NM_177986.2).

Original data:

1)- GenBank: AY177664.1 Submitted to GenBank in November 2002

2)- GenBank: AY227350.1 Submitted to GenBank in January 2003


It is important for independent scientists to determine how popular the human carry variant 2 of desmoglein 4.


Future research should emphasize on the role of desmoglein 4 in cell signaling that influence the development of skin, hair, and mucosa.



1- Nguyen,V.T. Chapter 19: “Experimental Mouse Model of Pemphigus Vulgaris: Passive Transfer of Nondesmoglein 1 and 3 Antibodies" (in) Chan L.S. (Ed.); ANIMAL MODELS OF HUMAN INFLAMMATORY SKIN DISEASES; CRC Press LLC, FL, USA (2003)

2. Kljuic A, Bazzi H, ..., Nguyen VT, Christiano AM. Desmoglein 4 in hair follicle differentiation and epidermal adhesion: evidence from inherited hypotrichosis and acquired pemphigus vulgaris. Cell. 2003 Apr 18;113(2):249-60. doi: 10.1016/s0092-8674(03)00273-3. PMID: 12705872.

Figure 3.

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9