Indeed, you will always have a difference between Nanodrop and gel concentrations, since you will always have variable amounts of nucleotides or small oligonucleotides coming from even little RNA degradation, small ncRNAs and fragmentation of contaminant genomic DNA after DNAse treatment. Nanodrop is useful to have an idea of the order of magnitude of the extracted RNA and its chemical purity (ratios 260/280 nm, 260/230 nm and shape of the absorbance curve), but you will ALWAYS need to check your RNA on gel for its integrity. That's what you see on your gel, two RNAs are heavily degraded as you realise yourself (1 and 3) and two RNAs are partially degraded (2, where the 18S and 28S are equal intensity, and 4, where the 18S is much more intense than the 28S rRNA, while you should have the 28S rRNA of double intensity than the 18S rRNA due to its more than double size and therefore more than double capacity to intercalate ethidium bromide, which means you lost the longer RNA species due to degradation - for example 28S in mouse is ca. 4700 nt, in human ca. 5000 nt).

As already stated, it is highly unsafe to use these degraded RNAs for quantification, because you introduce an uncontrollable variable, how much of gene expression you will be missing due to the degradation stochastically hitting your amplicon (even if amplicons are small in qPCR, it is still an uncertainty you don't have means to account for or measure). So, the solution is practice to extract better RNA, or search for a better extraction method, depending on what sample you are working with.

There are several possible extraction methods:

Trizol (or similar), based on total lysis of cells or tissues using a mixture of phenol and guanidine hydrocloride or isotiocyanate. This is well suited for dirty tissues (liver, kidney, intestine, stomach, spleen) or anyway samples where you extract large amounts of RNA (which might be your case, you do have a quite high RNA conc.) since it is difficult to get rid of the guanidinium due to its very low solubility in any medium except pure water and it will denature efficiently your RT if you don't dilute it sufficiently

Guanidinium isotiocyanate followed by CsCl cushion, which is an excellent total RNA extraction method, but you will have to handle the toxicity of Cs and will need an Ultracentrifuge, and will need high amounts of starting material.

Cytoplasmic extraction (triton-X based) followed by Proteinase K digestion and phenol/Chloroform extraction which is convenient for cleaner tissues and cells, works also with very low amounts of RNA,and carries on the least chemical contamination.

Columns (there are many, from many companies, using different principles), which come with different lysis buffers (in which is difficult to control degradation) and in my opinion are the worse method since the yield is always low (detaching the RNA from the column is quite inefficient) and all the small RNA species are lost.

Nanodrop measures nucleotide concentration, essentially, so yes: it doesn't differentiate between intact RNAs and degraded ones.

The assumption you make when making cDNA is that you have clean, good quality RNA, and while gel electrophoresis will tell you if your RNA is degraded (or degrading), it isn't a good method for determining RNA concentration.

However, if you have degraded RNA, you shouldn't be worrying about measuring the concentration accurately anyway, because you shouldn't use that RNA. Especially if you're planning on comparing expression levels between samples.

Is there any solution for this problem ?

Indeed, you will always have a difference between Nanodrop and gel concentrations, since you will always have variable amounts of nucleotides or small oligonucleotides coming from even little RNA degradation, small ncRNAs and fragmentation of contaminant genomic DNA after DNAse treatment. Nanodrop is useful to have an idea of the order of magnitude of the extracted RNA and its chemical purity (ratios 260/280 nm, 260/230 nm and shape of the absorbance curve), but you will ALWAYS need to check your RNA on gel for its integrity. That's what you see on your gel, two RNAs are heavily degraded as you realise yourself (1 and 3) and two RNAs are partially degraded (2, where the 18S and 28S are equal intensity, and 4, where the 18S is much more intense than the 28S rRNA, while you should have the 28S rRNA of double intensity than the 18S rRNA due to its more than double size and therefore more than double capacity to intercalate ethidium bromide, which means you lost the longer RNA species due to degradation - for example 28S in mouse is ca. 4700 nt, in human ca. 5000 nt).

As already stated, it is highly unsafe to use these degraded RNAs for quantification, because you introduce an uncontrollable variable, how much of gene expression you will be missing due to the degradation stochastically hitting your amplicon (even if amplicons are small in qPCR, it is still an uncertainty you don't have means to account for or measure). So, the solution is practice to extract better RNA, or search for a better extraction method, depending on what sample you are working with.

There are several possible extraction methods:

Trizol (or similar), based on total lysis of cells or tissues using a mixture of phenol and guanidine hydrocloride or isotiocyanate. This is well suited for dirty tissues (liver, kidney, intestine, stomach, spleen) or anyway samples where you extract large amounts of RNA (which might be your case, you do have a quite high RNA conc.) since it is difficult to get rid of the guanidinium due to its very low solubility in any medium except pure water and it will denature efficiently your RT if you don't dilute it sufficiently

Guanidinium isotiocyanate followed by CsCl cushion, which is an excellent total RNA extraction method, but you will have to handle the toxicity of Cs and will need an Ultracentrifuge, and will need high amounts of starting material.

Cytoplasmic extraction (triton-X based) followed by Proteinase K digestion and phenol/Chloroform extraction which is convenient for cleaner tissues and cells, works also with very low amounts of RNA,and carries on the least chemical contamination.

Columns (there are many, from many companies, using different principles), which come with different lysis buffers (in which is difficult to control degradation) and in my opinion are the worse method since the yield is always low (detaching the RNA from the column is quite inefficient) and all the small RNA species are lost.

Just use the relative band intensity (densitometry) and then use that information to normalize the concentrations of your four samples. That will give you equal amounts of the main band product in each reaction. However, the degraded product has the potential to interfere with your results as the guy above stated.

thank you all

1st. did you check purity using nanodrop. unless their A260/A280 values are equal, you can not compare the nanodrop cancentrations to compare them. As you said, the RNA contamination as well as G/C content effects the A260. and it seems that you have RNA contamination...alot. If you need to be very precise... 1st, get rid of RNAs, 2nd. use a UV/VIS spectrophotometer instead of nanodrop, 3rd. use a series of dilutions to measure the concentration. 4th. make sure their A260/A280 values are within the +-0.05 error margin. 5th. cross your fingers during recording :D

In addition to what the others have suggested above, traces of phenol or ethanol can affect the reading you get from nanodrop. Look for absorbence lower to 260nm.  Secondly, DNase1 treatment of RNA should eliminate any DNA that might affect the nano drop reading at A260. In general columns aren't efficient in extracting RNAs lower than 200 nt and might help reduce degradation products. 

Hi. In my case, I take amount of RNA (based on Nanodrop results) to get same concentration of cDNA after synthesis. The real concentration of cDNA might not be same between samples because of RNA degradation. However, I can make they balance by using internal control gene (for example r16S). After making the PCR of cDNA with internal control gene, I run gel, take picture and then use ImageJ program to compare the brightness of each band. Then, I can change the volume of my cDNA sample to make them same concentration. Good luck to you.

You could do this also if you are getting different concentration of RNA on nano-drop for example sample 1 contain = X and sample 2 contain =2x .

Just before c-DNA synthesis take equal amount .

you could do this by taking X amt of sample 1 and ½ of sample 2. Or you could take twice sample 1 so that your final amount is same i.e. 2x in both of your  RT –reaction.

hope this is useful

All the best